ABSTRACT Title of Thesis: THERMAL CYCLING DESIGN ...
-
Upload
khangminh22 -
Category
Documents
-
view
2 -
download
0
Transcript of ABSTRACT Title of Thesis: THERMAL CYCLING DESIGN ...
ABSTRACT
Title of Thesis: THERMAL CYCLING DESIGN ALTERNATIVES FOR THE
POLYMERASE CHAIN REACTION
Degree Candidate: Monte Lewis
Degree and Year: Master of Science, 2005
Thesis directed by: Associate Professor Keith E. Herold Department of Mechanical Engineering
PCR (polymerase chain reaction) is a process by which a small amount of DNA is
amplified many times to yield an easily detectable amount. This process is widely used
for the detection of bacterial pathogens for biodefense, in basic research, criminal
identification, and disease detection in humans. The reaction mixture must be cycled
repeatedly between three different temperature levels in PCR. The reaction mixture is
first heated at 94 °C, cooled to 54 °C, and then heated to 72 °C. This cycle is repeated
20-40 times.
The main objective of the work reported here is to evaluate alternative
heating/cooling schemes for PCR with the ultimate goal of speeding up a PCR reaction.
A secondary goal is to arrive at a design that is consistent with battery operation to allow
for a portable PCR device. Insight is gained about interactions between the PCR reaction
and the engineering system.
The first device analyzed for use as a thermal cycler is a Peltier thermoelectric
cooler. A single thermoelectric cooler is characterized. The maximum and average
heating rates and power consumption are determined through experiment. The
thermoelectric cooler is evaluated under different conditions and modeled with the
lumped capacitance method to reveal the characteristics of the device as a thermal cycler.
This Peltier device is used as a heating/cooling system for subsequent work on alternative
reaction vessels. A commercially available Peltier thermocycler is also analyzed to
determine the heating and cooling rates and power consumption. PCR results are
presented that show the time limits of the PCR reaction.
The second device analyzed for use as a thermal cycler is a convective
thermocycler. An advantage of the convective thermocycler is it allows for a wide range
of vessel geometries. Two variants of convective thermocyclers were built and tested for
performance. Heat transfer to a sample is modeled using finite element methods as well
as with the lumped capacitance method and corroborated with experiment. Reaction
vessels of different sizes and material are modeled in order to better understand the
relationship between the air velocity and heating characteristics of the thermal cycler and
the reaction volume.
The third device analyzed is a microheater system. The microheater design is
arrived at as the preferred design for high speed PCR. Microheaters were fabricated
using standard MEMS techniques. Problems encountered in the design, fabrication, and
use of these heaters are presented. The microheaters are evaluated for thermal cycling
performance under different conditions. Chambers fabricated out of polycarbonate are
tested for use as reaction volumes. Two dimensional finite element analysis is used to
model these systems and the models are corroborated with experiment. The behavior of
liquid in polycarbonate chambers undergoing thermal cycling is investigated. Problems
associated with thermal cycling in polycarbonate chambers is discussed along with
possible solutions.
THERMAL CYCLING DESIGN ALTERNATIVES FOR THE POLYMERASE CHAIN
REACTION
by
Monte Lewis
Dissertation submitted to the Faculty of the Graduate School of the University of Maryland, College Park in partial fulfillment
of the requirements for the degree of Master of Science
2005 Advisory Committee: Associate Professor Keith E. Herold, Chair Associate Professor Elisabeth Smela Assistant Professor Bao Yang
ii
ACKNOWLEDGMENTS
I would like to thank my professor, Dr. Keith Herold, for giving me the opportunity to work for him, for his patience, and for his help in completing this project. I am very thankful to Dr. Jungho Kim for his help on the control system for the microheaters. The USDA funded this work, for which I am grateful. I was very lucky to have worked with our post doc from Ukraine, Andrey Matviyenko who spent many days doing PCR and working on mechanisms for PCR. Nikolai Sergeev and Avi Rasooly also provided valuable input to the project that I am thankful for. I would like to say thank you to Tom Laughran who taught me many things, was always very helpful and responsive, and set an excellent example for cleanroom behavior. Tom also graciously deposited metal on our samples. Nolan Balew and Jon Hummel also provided valuable support in the clean room. I would like to thank the many people I have met in the clean room and provided me with information including Steve Fanning, Samuel Moseley, and Remi Delille. Sayeed Moghaddam and Vytenis Benetis went out of their way to help me in the cleanroom and with the microheater design. I am also thankful to Dr. Elisabeth Smela who allowed the use of her lab equipment and was also an excellent teacher. I would like to thank Lou Hromalo for helping me in several situations. Corey Skurdal also provided valuable input and performed PCR many times. The Terrapin Masters swim team provided me with much needed exercise and camaraderie during my time here for which I am thankful.
iii
TABLE OF CONTENTS
LIST OF TABLES ........................................................................................................... vi
LIST OF FIGURES ........................................................................................................ vii
SYMBOLS...................................................................................................................... xiii
1. INTRODUCTION TO PCR AND THERMOCYCLERS.....................................1
1.2 PCR reaction and detection reagents ........................................................5
1.3 Thermocycler metrics ...............................................................................7
2. TESTING OF PELTIER THERMOCYCLER FOR PCR.................................10
2.1 Testing of an independent TEC, apparatus and procedures....................10
2.2 Independent TEC testing results .............................................................13
2.2.1 Heating flux characteristics of TEC........................................................18
2.2.2 Cooling flux characteristics of TEC .......................................................26
2.2.3 Power consumption.................................................................................29
2.3 Independent TEC testing discussion.......................................................29
2.4 Testing of a commercially available Peltier thermocycler,
apparatus and procedures........................................................................31
2.4.1 Commercial Peltier thermocycler testing results ....................................33
2.4.2 Power consumption of the commercial device ......................................34
2.4.3 PCR results with the commercial device ................................................35
3. ANALYSIS OF CONVECTIVE THERMOCYCLERS FOR PCR...................42
3.1 Testing of a convective thermocycler, apparatus and procedures ..........42
3.2 Convective thermocycler test results ......................................................45
3.2.1 Power consumption of convective thermocyclers ..................................49
iv
3.2.2 Convection characteristics of the thermocycler......................................50
3.2.3 System response with a test sample ........................................................55
3.3 Convective thermocycler options to enhance performance ....................66
4. ANALYSIS OF MICROHEATERS FOR PCR...................................................77
4.1 Synopsis of work done on microheaters by other groups .......................77
4.1.1 Comparison of designs............................................................................82
4.2 Microheater design .................................................................................85
4.2.1 Electronic control system design ............................................................95
4.3 Microheater fabrication...........................................................................99
4.3.1 Microheater fabrication equipment and techniques................................99
4.3.2 Microheater fabrication sequence .........................................................101
4.3.3 Microheater fabrication results .............................................................103
4.4 Microheater heat treatment and calibration ..........................................106
4.5 Microheater testing ...............................................................................110
4.5.1 Microheater testing results....................................................................112
5. MICROCHAMBERS FOR PCR ........................................................................116
5.1 Microchamber design considerations ...................................................124
5.2 Microchamber fabrication equipment and techniques..........................134
5.3 Microchamber testing ...........................................................................136
5.3.1 Countersunk polycarbonate chambers ..................................................137
5.3.2 Glass capillaries ....................................................................................139
5.3.3 Thermally bonded polycarbonate chambers .........................................140
5.3.3.1 Thermally bonded polycarbonate chamber sealing ..............................145
v
6. SUMMARY, CONCLUSIONS AND FUTURE WORK.......................................149
6.1 Summary ...............................................................................................149
6.1.1 Testing of Peltier thermocyclers ...........................................................150
6.1.2 Testing of convective thermocyclers ....................................................150
6.1.3 Testing of a microheater thermocycler .................................................151
6.2 Conclusions...........................................................................................152
6.3 Future work...........................................................................................157
REFERENCES ...............................................................................................................159
vi
LIST OF TABLES
Table 1.1 Primer sequences used in amplification of B. anthracis ....................................6 Table 2.1 Properties of aluminum oxide used in the analysis..........................................18 Table 2.2 Comparison of TEC with and without cooling fan ..........................................29 Table 2.4 Protocol for elongation time experiment..........................................................35 Table 3.1 Results of air velocity analysis for single and double fan convective
thermocyclers...................................................................................................53 Table 3.2 Nusselt number and convection coefficients for single and double fan
convective thermocycler ..................................................................................55 Table 3.3 Calculated time constants (τ) for the air velocities within the single and double fan convective thermocycler for a modeled 200 µL polypropylene tube with 10 µL of water .........................................................57 Table 3.4 Polypropylene properties .................................................................................58 Table 3.5 Thermal conductivities of candidate materials for PCR reaction vessels ..............................................................................................................73 Table 3.6 Biot number for varying polypropylene cylindrical diameters ........................74 Table 4.1 Comparison of thermocycling devices.............................................................84 Table 4.2 Result of heat treatment on heater and temperature probe.............................107 Table 4.3 Comparison of microheater operation with and without heat sink ................115 Table 5.1 Model to predict pressure in sealed two-phase system ..................................131 Table 6.1 Comparison of thermocyclers ........................................................................149
vii
LIST OF FIGURES
Figure 1.1 One cycle of the PCR process.........................................................................2 Figure 1.2 A PCR cycle showing plateaus at 54, 72 and 94 °C .......................................2 Figure 1.3 Schematic of PCR cycle with dashed areas representing times included in the average power over cycle calculation ....................................8 Figure 2.1 Thermoelectric thermocycler setup...............................................................10 Figure 2.2 H-bridge configuration used in the control of the TEC ................................11 Figure 2.3 Side view of TEC setup ................................................................................12 Figure 2.4 Simulated PCR cycle for TEC without cooling fan , 13.8 V input...............14 Figure 2.5 Simulated PCR cycle for TEC with cooling fan on, 13.8V input.................14 Figure 2.6 Steady state heating mode of the TEC without cooling fan and 10.0 V input ..................................................................................................15 Figure 2.7 Steady state heating mode of the TEC with cooling fan and 10.0 V input ..................................................................................................16 Figure 2.8 Steady state cooling mode of the TEC without cooling fan and 13.8 V input............................................................................................16 Figure 2.9 Steady state cooling mode of the TEC with cooling fan and 13.8 V input ..................................................................................................17 Figure 2.10 Heat flux produced by thermoelectric cooler during heating from 72 to 94 °C and corresponding top and bottom of TEC temperature with 13.8 V input, cooling fan ..................................................20 Figure 2.11 Calculated Performance of TEC using hot and cold side temperatures from experiment ......................................................................21 Figure 2.12 2-D geometry used for cooling and heating model.......................................23 Figure 2.13 Calculated temperature for top of the TEC superimposed on graph of actual temperature for top of TEC for 13.8 V input and cooling fan on.........................................................................................23 Figure 2.14 Heat flux produced by thermoelectric cooler during heating from 72 to 94 °C with 13.8 V input, cooling fan off ....................................24
viii
Figure 2.15 Calculated temperature from 2-D model for top of the TEC superimposed on graph of actual temperature for top of TEC for 13.8 V input and cooling fan off ................................................25 Figure 2.16 Cooling flux produced by thermoelectric cooler during cooling from 94 to 54 °C, 13.8 V Input, cooling fan on ...............................26 Figure 2.17 2-D calculated temperature for top of the TEC superimposed on graph of actual temperature for top of TEC for 13.8 V input and cooling fan on ...............................................................................................27 Figure 2.18 Cooling flux produced by TEC during cooling from 94 to 54 °C, 13.8 V input, cooling fan off ....................................................28 Figure 2.19 Techne thermocycler analyzed in this section ..............................................31 Figure 2.20 Aluminum heating block mounted on top of TECs in Techne thermocycler ......................................................................32 Figure 2.21 Simulated PCR cycle using commercial Peltier thermocycler .....................33 Figure 2.22 PCR reaction results for extension time test, 40 s elongation time ..............36 Figure 2.23 PCR reaction results for extension time test, 30 s elongation time ..............37 Figure 2.24 PCR reaction results for extension time test, 25 s elongation time ..............38 Figure 2.25 PCR reaction results for extension time test, 30 s elongation time with 15 s at denaturation and annealing steps.....................39 Figure 2.26 Glass capillaries run with a 1/4000 volumetric concentration of ethidium bromide......................................................................................40
Figure 3.1 Convective thermocycler setup #1................................................................42
Figure 3.2 Heater mounted within vent pipe..................................................................43 Figure 3.3 Double fan convective thermocycler setup...................................................43 Figure 3.4 Heater element within double fan convective thermocycle..........................44 Figure 3.5 Simulated PCR cycle using the single fan convective thermocycler............46 Figure 3.6 Simulated PCR cycle using the double fan convective thermocycler ..........46
ix
Figure 3.7 Cross section of single fan convective thermocycler showing mixing walls located within vent pipe............................................52 Figure 3.8 Radial temperature measurements for single fan convective thermocycler with heater and fan at duty cyle = 1........................................48 Figure 3.9 Schematic of control volume used for air velocity determination................50 Figure 3.10 200 µL PCR tube with 10 µL of water used in convective analysis.............55 Figure 3.11 2-D model used for convective analysis of 200 µL polypropylene tube filled with 10 µL of water .....................................................................58 Figure 3.12 Experimental results for single fan convective thermocycler with 200 µL tube filled with 10 µL of water placed inside vent pipe ..........59 Figure 3.13 2-D model correlation with experimental results for single fan convective thermocycler with 200 µL PCR tube containing 10 µL of water during heating from 52 – 94 °C ...........................................60 Figure 3.14 2-D model correlation with experimental results for single fan convective thermocycler with 200 µL PCR tube during cooling from 94 – 52 °C ...............................................................................61 Figure 3.15 Experimental results for double fan convective thermocycler with 200 µL polypropylene tube filled with 10 µL of water placed inside vent pipe..................................................................................62 Figure 3.16 Lumped capacitance model with experimental results for double fan convective thermocycler with 200 µL PCR tube during heating from 54–94 °C ..............................................................63 Figure 3.17 Lumped capacitance model with experimental results for double fan convective thermocycler with 200 µL PCR tube during cooling from 94–54 °C .................................................................................65 Figure 3.18 Plot of time constant versus air velocity for cylindrical configuration with constant diameter, length, wall thickness, surface temperature, and air temperature......................................................67 Figure 3.19 Plot of time constant versus diameter for cylindrical polypropylene configuration ........................................................................68 Figure 3.20 Lumped capacitance model of reaction volume temperature for glass capillary of 890 µm outside diameter with 5 µL of water inside ...................................................................................................70
x
Figure 3.21 Plot of time constant versus diameter for spherical configuration with constant air velocity, and wall thickness ..............................................72 Figure 3.22 Biot number versus wall thickness for a poly propylene capillary of outside diameter 890 µm ...........................................................75 Figure 3.23 Biot number versus wall thickness for a glass capillary of outside diameter
890 µm ..........................................................................................................76 Figure 4.1 Heater with integrated well as designed by El-Ali et al................................78 Figure 4.2 PCR reaction vessel designed by Zhao et al. ................................................80 Figure 4.3 Microheaters with integrated reaction vessels as designed by Zou et al. .....81 Figure 4.4 Flip chip design process used by Zou et al. for fabrication of PCR
microheaters..................................................................................................82 Figure 4.5 Schematic of 2-D model used in analysis of heating power.........................86 Figure 4.6 Response of 3.2 mm diameter polycarbonate chamber of thickness 2 mm
with 10 µL of water in it and a heat flux of 20 W/cm2 ................................86 Figure 4.7 Response of 5 mm diameter polycarbonate chamber of depth 2 mm with a 10 µL sample of water in it and a heat flux of 20 W/cm2..................88 Figure 4.8 Schematic of 5 mm heater.............................................................................91 Figure 4.9 Exploded view of 5 mm heater with some elements deleted for clarity.......92 Figure 4.10 Schematic of 2.5 mm heater..........................................................................93 Figure 4.11 Exploded view of 2.5 mm heater ..................................................................94 Figure 4.12 Gold lead mask for 5 mm heater...................................................................95 Figure 4.13 Gold lead mask for 2.5 mm heater................................................................95 Figure 4.14 Circuit schematic for microheater control ....................................................96 Figure 4.15 Schematic of circuit used to measure temperature of platinum temperature sensor ........................................................................................98 Figure 4.16 Simplified fabrication sequence for microheaters ......................................101 Figure 4.17 Effect of chlorobenzene on photoresist development.................................102
xi
Figure 4.18 Delamination of heater elements from soda lime glass substrate ...............104 Figure 4.19 Heater elements from a heater with no discontinuities...............................105 Figure 4.20 Comparison of 30 µm heater elements with 50 µm temperature sensor ....105 Figure 4.21 Points of resistance measurement for 5 mm microheater ...........................107 Figure 4.22 Temperature versus resistance with linear fit for microheater....................108 Figure 4.23 Temperature versus resistance graph with linear fit for microheater
temperature sensor ......................................................................................109 Figure 4.24 Comparison of initial calibration of temperature sensor and after 1 week of
use ...............................................................................................................110 Figure 4.25 Microheater test fixture...............................................................................111 Figure 4.26 Microheater test fixture with microheater mounted on heat sink ...............112 Figure 4.27 Heating and cooling characteristics of microheater with cooling fan, no heat sink, 12 V input ..............................................................................113 Figure 4.28 Heating and cooling characteristics of microheater with cooling fan, heat sink, 12 V input ...................................................................................114 Figure 5.1 Silicon reaction chamber fabricated by Daniel et al. ..................................116 Figure 5.2 Simplified fabrication sequence used by Daniel et al. ................................117 Figure 5.3 Polycarbonate chamber fabricated by Yang et al. ......................................120 Figure 5.4 2-D heat transfer model for polycarbonate chamber ..................................124 Figure 5.5 2-D conduction model of 5 mm square polycarbonate chamber on 5 mm
square microheater ......................................................................................125 Figure 5.6 2-D conduction model of 3 mm square polycarbonate chamber on 5 mm
square microheater ......................................................................................126 Figure 5.7 2-D conduction model of 3 mm square silicon chamber on 5 mm square
microheater .................................................................................................127 Figure 5.8 Schematic of 800 µm capillary on microheater ..........................................129
xii
Figure 5.9 Polycarbonate reaction vessel thermally bonded from three 250 µm thick sheets...........................................................................................................130
Figure 5.10 Predicted pressure rise inside closed reaction vessel with water and 33% of
the volume filled with air............................................................................132 Figure 5.11 Predicted bubble volume inside closed reaction vessel with water and 33%
of the volume filled with air........................................................................133 Figure 5.12 Polycarbonate chamber fabricated from three 250 µm thick sections........134 Figure 5.13 Countersunk polycarbonate chambers created with router bit....................135 Figure 5.14 Bubble formation in untreated polycarbonate chambers covered with
mineral oil ...................................................................................................137 Figure 5.15 2 mm thick polycarbontate after O2 plasma etch at A)100 W, B) 200 W, C)
250 W..........................................................................................................138 Figure 5.16 PCR cycle showing 5 mm square microfabricated heater and temperature
inside a 400 µm inside diameter glass capillary filled with ethylene glycol ............................................................................................140 Figure 5.17 PCR cycle showing 5 mm square microfabricated heater and temperature
inside a 5 mm square polycarbonate reaction volume filled with ethylene glycol....................................................................................141 Figure 5.18 Simulated PCR cycle showing 5 mm square microfabricated heater and
temperature inside a 3 mm square polycarbonate reaction volume filled with ethylene glycol ............................................................................................142
Figure 5.19 Simulated PCR cycle showing 5 mm square microfabricated heater and
temperature inside a 3 mm square polycarbonate reaction volume with a 125 µm thick top and bottom section filled with ethylene glycol .....................144
Figure 5.20 Simulated PCR cycle showing 5 mm square microfabricated heater and
temperature inside a 3 mm square polycarbonate reaction volume with a 125 µm thick bottom section and a 250 µm thick top section filled with ethylene glycol...........................................................................................................145
xiii
SYMBOLS
A area, m2 Bi Biot number
c specific heat, J/kg K cp specific heat
D diameter, m at constant pressure, J/kg K
f friction factor; fraction G area/length of
h convection coefficient, W/m2 K; thermoelectric device (cm)
enthalpy, J/kg
I electrical current, A k thermal conductivity, W/m K
L length, m m mass, kg
N number Nu Nusselt number
P power, W; Pressure, N/m2 Pr Prandtl number
q heat transfer rate, W "q heat flux, W/m2
R electrical resistance, Ω; universal gas Re Reynolds number
constant divided by the molecular
weight, kJ/kg K
T temperature,°C, K t time, s; thickness, m
V voltage, V; volume, m3; Velocity, m/s W percent on time; work, J
w width, m x displacement, m; quality
Greek letters
α Seebeck coefficient, V/K ∆ change in
υ specific volume, m3/kg μ viscosity, kg/s m
ρ mass density, kg/m3; electrical τ time constant, s
resistivity, Ω-m
xiv
Subscripts
air refers to air ave average
B beginning c during cooling; cold
D diameter side; characteristic
E end h during heating, hot side
i initial condition; inlet condition l liquid
max maximum min minimum
n point n o outlet condition
P parallel p polypropylene
RT thermometer resistance S series
s surface; solid t thermal; thermometer
tot total v vapor
w water wl liquid water
wv water vapor ∞ free stream conditions
Overmarks
. per unit time, /s surface average conditions,
time mean
1
1. INTRODUCTION TO PCR AND THERMOCYCLERS
PCR is a process by which a small amount of DNA is amplified many times to
yield an easily detectable amount. This process is widely used for the detection of
bacterial pathogens for biodefense (Draghici, Khatri et al. 2005; Hindson, Makarewicz et
al. 2005). PCR is also utilized in other applications such as basic research, criminal
identification, and disease detection in humans (Aryee, Bailey et al. 2005; Gill 2005).
DNA, primers that dictate what part of the DNA is to be copied, nucleotides that form the
building blocks for copying, buffer, polymerase that is the machine for copying, and
magnesium ions are needed for the reaction (Cadoret, Rousseau et al. 2005). Three
temperature levels are required in a typical PCR reaction. The first step involves heating
to around 94 °C. The double stranded DNA is denatured, or split into two single strands
during this step. The second step involves cooling to approximately 54°C to allow the
primers to attach to the single-stranded DNA. The primers are what determine which
section of DNA will be copied. They are designed such that they straddle a
predetermined specific site on the DNA based on the sequence of the DNA. The third
step is performed at around 72 °C. The actual copying of the DNA is performed by the
DNA polymerase during this step. The DNA polymerase attaches to the primer site and
copies DNA using the nucleotides that are in solution as building blocks. This cycle is
repeated approximately 30 times (Nam, Srinivasan et al. 2005). The steps in the PCR
cycle are illustrated graphically in Figure 1.1. The temperature profile of a PCR cycle is
shown in Figure 1.2.
2
Figure 1.1 One cycle of the PCR process (Vierstraete 2004)
45
50
55
60
65
70
75
80
85
90
95
100
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180
Time (seconds)
Tem
pera
ture
(°C
)
Figure 1.2 A PCR cycle showing plateaus at 54, 72, and 94 °C
3
A typical PCR reaction in a commercial thermocycler takes between 30 minutes
(LightCycler) and 2 hours (typical Peltier systems). This time duration is acceptable for
some applications, however, many other applications exist where a rapid reaction is
desirable such as clinical applications where treatment depends on the result. The
primary objective of this thesis was to investigate system alternatives for high speed
PCR. A secondary objective was to minimize power consumption to allow eventual
portable operation with battery power. A series of alternative designs were tested in this
effort and each was found to have a set of unique and useful characteristics.
The cycle time can vary based on the situation. The cycle time is strongly
dependent upon the characteristics of the thermocycler. If the thermocycler takes a long
time to reach the setpoint, then this must be accommodated. If heat conduction or
convection is slow through the reaction chamber in spite of a fast heater, then the cycle
time would also be longer. PCR reactions run in this lab in commercially available
Peltier type thermocyclers are usually run for two hours. This type of thermocycler has
an aluminum heating block that weighs 41 g. This block takes time to heat up and slows
the heating of the actual reaction volume. The commercially available convective
LightCycler has been reported to produce PCR results in as little as 20 minutes (Nitsche,
Steur et al. 1999). One reason PCR occurs so rapidly in this device is that the reaction
volume heats up very quickly. Another factor that influences the speed of a PCR reaction
is the length of DNA strand that is to be copied. Shorter strands take less time to copy
then longer strands. The insertion rate for nucleotides during primer extension (72 °C) is
50 base pairs /s (Scheinert, Behrens et al. 2005). A base pair (bp) is simply one link
between the two single strands of DNA. If the DNA fragment to be copied is only 100
4
bp then it would require only two seconds at 72 °C to be copied. If the DNA fragment
was 400 bp then it would require 8 seconds to be copied. The type of DNA polymerase
that is used can significantly alter the reaction time. Hot Start TAQ DNA polymerase
requires 15 minutes of heating at 95 °C to become effective. This type of polymerase is
commonly used in PCR reactions because it reduces the occurrence of nonspecific
amplicons (Kermekchiev, Tzekov et al. 2003). The term amplicon refers to the DNA
segments produced from the PCR reaction. Other types of DNA polymerase may not be
as specific, but they do not require this 15 minute heating time. In conclusion, the cycle
time is influenced by the thermocycler, the reaction volume, the length of the DNA
segment to be copied, and the type of DNA polymerase used.
The amplified DNA may be detected in a number of ways. One way is by taking
the sample and moving it through a gel with the application of an electric field. Smaller
DNA segments migrate through the gel more quickly while larger DNA fragments move
more slowly. The DNA fragment is detected by determining how far it migrates through
the gel. The distance migrated through the gel is revealed by illuminating the DNA under
a UV light source. The UV light causes fluorescent dye that is added to the DNA to
fluoresce, thus enabling one to see the location of the DNA fragment. Another method of
detection involves the use of probes that have the complementary sequence to the single
strand of DNA that is being copied. If probe locations are occupied with double stranded
DNA, then it is known that PCR occurred on that particular piece of DNA (Lane,
Everman et al. 2004).
PCR requires that the reaction volume undergo changes in temperature, therefore
a thermal cycler is needed to perform this operation. Thermal cyclers come in a variety
5
of shapes and sizes, but generally may be broken down into two branches: convective
thermocyclers and conductive thermal cyclers with other types being reported such as
microwave assisted (Fermer, Nilsson et al. 2003) and infrared (Lee, Tsai et al. 2004).
Convective thermocyclers operate on the principal of flowing air past a reaction vessel in
order to increase or decrease the temperature of the reaction vessel. These units consist
of a reaction chamber, housing, heater, fan, and control system (Wittwer, Fillmore et al.
1989). The other type of thermocycler relies on conduction from a heated plate to drive
the reaction. The heated plate could take on a variety of shapes and sizes varying from a
thermoelectric cooler to a microfabricated heater. The purpose of this thesis is to
investigate the properties of these different types of thermocyclers.
1.2 PCR reaction and detection reagents
PCR was run many times to determine the efficacy of the thermocyclers tested.
Several reagents are necessary to run the reaction. The DNA was genomic DNA from
non-infectious B. anthracis (Sterne strain) which is the strain used to make vaccine. Hot
start TAQ DNA polymerase, 10 X buffer solution, and 25 mM MgCl2 solution supplied
by Qiagen were used in all PCR runs. The dNTPs, which are the building blocks the
DNA polymerase uses to copy the target DNA, were obtained as a 10 mM solution from
Fisher Scientific. Primers that determine which section of DNA is to be copied were
designed and ordered from Operon. The gene amplified was the piplc gene of B.
anthacis. Table 1.1 shows the exact sequence of the primers used and the amplicon
length that was expected based on the designated primers.
6
Table 1.1 Primer sequences used in amplification of B. anthracis
Primer Sequence Amplicon length piplc-R ATA TTT ATC ACA TTT GTA TTT GCT TTA CAT GA 879 piplc-F CTT CTT GAT ACA ATA ATG GTG ACC ACT piplc-R2 ATG AAA CGA TGA ACA ATA GTG AGG A 216
The R and F suffixes at the end of the primer name in the first column of Table 1.1
designate whether the primer was a forward primer or a reverse primer. Forward primers
attach to one end of the single stranded DNA segment to be copied while reverse primers
attach to the other end of the complementary strand. Copying proceeds as shown in
Figure 1.1. The first two primers listed in Table 1.1 were used to amplify a segment of
DNA 879 bp long. The third row of Table 1.1 is the reverse primer that was used with
the forward primer to produce an amplicon of length 216 bp. Distilled water was used in
all reactions. Bovine serum albumin was also added to the reactions. Bovine serum
albumin helps to coat the reaction vessel and stop protein adsorption to the surface (Liu,
Rauch et al. 2002).
One method used to detect the PCR products was gel electrophoresis. A QS710
horizontal gel electrophoresis unit was obtained from Fisher Scientific. The power
supply used with the gel electrophoresis unit was a Buchler DC power supply, model
31153. A Kodak EDA 290 camera stand was mounted on top of a Foto Prep I by
Fotodyne UV light source to take pictures of the gel. Kodak 1D version 3.6 was the
software used to take the pictures. Omni Pur PCR plus agarose was used to separate the
PCR products. The buffer solution used in the electrophoresis unit was TBE Omni Pur
10 X powder. Ethidium bromide was used to attach to the double stranded DNA during
gel electrophoresis. Ethidium bromide fluoresces under UV light when attached to
7
double stranded DNA. Exact gene 100 bp PCR DNA ladder was used to determine the
length of PCR amplicons.
Another method used to detect the occurrence of PCR was by adding ethidium
bromide to the PCR solution at a concentration of 1:4000 by volume. This method
worked well for illuminating PCR product in glass capillaries and sped up the PCR
process by skipping the time consuming gel electrophoresis step.
1.3 Thermocycler metrics
Metrics that will be used to discuss the performance of the thermocyclers are
heating rate, cooling rate, average heating rate, average cooling rate, and power
consumption. The maximum heating rate is defined as the maximum rate of change of
temperature when the heater is active during heating within the temperature range of a
typical PCR reaction (54 to 94 °C). The heating rate is based on the surface temperature
of the heater, or in the case of the convective thermocycler, the air temperature.
max
.
dtdTT h = (1.1)
where T is the temperature and t represents time. The cooling rate is defined as the
maximum rate of decrease in temperature during cooling from 94 to 54 °C.
min
.
dtdTT c = (1.2)
The average cooling and heating rate are defined as follows
12
12_
ttTTTh −
−= ,
12
12_
ttTTTc −
−= (1.3)
8
Figure 1.3 Schematic of PCR cycle with dashed
areas representing times included in the average
power over cycle calculation
The average rates are influenced by the control system implemented to control the
temperature of the heater, unlike the maximum heating and cooling rates, which are
maximum values calculated when the heater was less inhibited by the control system.
The average heating rate would be much faster if the heater was on at full power during
the entire interval between T2 and T1. This is not the case, however, because the control
system intervenes before the heater temperature reaches the set point in order to avoid
overshoot.
The energy consumption can be reported in several ways. The maximum energy
consumption per unit time during heating and cooling is one of those ways. During the
heating stages where the PCR reaction increases in temperature, the heater is on at full
power. This means that the power required is given by
VIP = (1.4)
where P is the power, V is the voltage, and I is current draw of the device.
The power requirement
varies throughout the cycle, so the
average power over a cycle can be
used to characterize the system. The
average power over a cycle includes
the time required for heating and
cooling as well as five seconds of
each plateau period as shown in
Figure 1.3.
9
The average power is defined in this way in order to compare different heaters.
All of the heaters tested had at least a five second plateau period, so this method would
compare them with the same standard. The average power over a cycle could be
increased or decreased if the energy of a very long plateau was included in the
calculation. The average power over a cycle is given by the following equation
tPN
P n
N
cycleave ∑=0
/1 (1.5)
where Pave/cycle is the average power over a cycle, N is the number of seconds in the cycle,
t is the time of each period, and Pn is the power used by the system during period n.
Some of the systems described in this thesis were controlled using a pulse width
modulation scheme where the power was applied during some fraction of each control
period (period used here was 1 sec). This control scheme is referred to as pulse width
modulation (PWM). In this case, Pn is given by
VIWPn = (1.6)
where W is the time per period that the heater was on.
10
2. TESTING OF PELTIER THERMOCYCLER FOR PCR
A thermoelectric cooler (TEC) was tested to determine its efficacy for performing
PCR. The thermoelectric cooler was mounted on an aluminum heat sink with an attached
cooling fan. The TEC was tested for power consumption, heating and cooling rate, and
heating and cooling flux. A commercially available thermocycler using four
thermoelectric coolers with an aluminum block to hold reaction vessels was also tested to
determine the same properties.
2.1 Testing of an independent TEC, apparatus and procedures
A single TEC was tested using the setup shown in Figure 2.1. A Mastech power
supply was used as the voltage source. A Type T thermocouple was used to monitor the
temperature of the surface of the heater. The InstruNet was used as the data acquisition
and controller for the setup.
Figure 2.1 Thermoelectric thermocycler setup
InstruNet model 100 analog/digital input/output system
Omega solid state relay part no. SSRDC100VDC12
Melcor thermoelectric cooler, model HT2-12-30-T2
(50 x 50 x 7) mm finned heat sink with 12 VDC fan
Copper-constantan (type T) thermocouple
Mastech DC power supply part no. HY1803D
11
The thermoelectric cooler (TEC) was controlled with a Visual Basic program
interfaced with the InstruNet model 100 analog/digital input/output system. The TEC
was controlled with a pulse width modulation scheme with a pulse frequency of 1 Hz.
The program calculated the appropriate pulse width and controlled the relays accordingly
via the InstruNet interface. Four solid state relays were utilized in an H-bridge
configuration allowing the direction of the current to be switched based on whether the
TEC was in heating or cooling mode as shown in Figure 2.2.
Figure 2.2 H-bridge configuration used in the control of the TEC. The components
labeled SSRi are solid state relays. Arrow on figure shows current flow in heating mode.
A forward current flows through the TEC when SSR1 and SSR3 are opened via a signal
from InstruNet (SSR2 and SSR4 remain closed). This is the heating mode of operation.
The TEC was put in cooling mode by closing SSR1 and SSR3 and opening SSR2 and
SSR4. This configuration allowed current to flow through the TEC in the opposite
direction.
12
The thermoelectric cooler was placed on the heat sink as shown in Figure 2.3.
The bottom of the TEC was coated with synthetic grease to aid in heat transfer between
the heat sink and the TEC. A glass slide, representing an in situ PCR configuration, was
mounted on top of the TEC and held in place with two metal strips. The thermocouple
was attached to the slide and held in place with a piece of tape. The purpose of this
thermocouple was to control the temperature of the slide. Another thermocouple was
taped to the heat sink. In some tests, thermocouples were attached to the top and bottom
surfaces of the TEC with glue in order to monitor the temperature of those two surfaces.
Voltage and current readings were obtained from the digital display on the power supply
(Mastech HY1803D). The TEC was tested over the range 9 – 13.8 VDC to assess the
effect on power consumption.
Figure 2.3 Side view of TEC setup
The data acquisition program recorded data once per second. Data recorded
included the temperature of the slide and top of TEC, the pulse width, time, and set point
TEC
Cooling fan
Heat sink
Glass slide
Metal compression strip
Compression screw Grease between slide and TEC and between TEC and heat sink
13
temperature (set point temperature varied during the run according to a programmed
input function). Using the recorded data, heating rate, cooling rate, and power
consumption were calculated. Only the power used by the heater was included in the
calculations for power consumption. Power consumed by the heat sink cooling fan, the
computer, and power supply were not included. The heating and cooling rates were
calculated as described in the thermocycler metrics section.
2.2 Independent TEC testing results
The TEC was tested with two different configurations to determine its behavior
under different conditions. The first condition was with the cooling fan on the heat sink
on. The fan was run continuously and was never turned off. The second condition was
with the fan on the heat sink off at all times.
The maximum heating rate for the TEC was 5.78 K/s with an input voltage of
13.8 V without the heat sink fan turned on as shown in Figure 2.4. The top of TEC curve
in Figure 2.4 also shows that the average heating rate for heating between 72 and 94 °C
was 2.0 K/s while the average heating rate between 54 and 72 °C was 1.4 K/s. The
maximum cooling rate for this setup was 5.2 K/s. The average cooling rate between 94
and 54 °C was 4.3 K/s.
14
253035404550556065707580859095
100
0 15 30 45 60 75 90 105 120 135 150 165 180
Time (seconds)
Tem
pera
ture
(°C
)
SetpointTop of TECBottom of TEC
Figure 2.4 Simulated PCR cycle for TEC without cooling fan , 13.8 V input
The TEC was next tested with the cooling fan on the heat sink on. The top of
TEC trace in Figure 2.5 was used to calculate the maximum heating and cooling rates as
well as the average rates. The maximum heating rate was 5.2 K/s. The average heating
rate between 54 and 72 °C was 1.8 K/s while the average heating rate between 72 and
94 °C was 1.3 K/s. The maximum cooling rate of the TEC with cooling fan was 6.6 K/s.
The average cooling rate was 5.8 K/s.
253035404550556065707580859095
100
0 15 30 45 60 75 90 105 120 135 150 165 180Time (seconds)
Tem
pera
ture
(°C
)
SetpointTop of TECBottom of TEC
Figure 2.5 Simulated PCR cycle for TEC with cooling fan on, 13.8V input
15
The TEC was next tested to determine its behavior at steady state. The TEC was
allowed to reach steady state while receiving full power (duty cycle = 1) and the results
were recorded. The TEC was powered with 10 V during this testing. The results for the
case where the TEC was allowed to reach steady state in heating mode with no cooling
fan are shown in Figure 2.6. Heating mode means that the top of the TEC was hot while
the side against the heat sink was cold.
Figure 2.6 Steady state heating mode of the TEC without cooling fan and 10.0 V input
The final temperature differential between hot and cold side was 55 K for the case
where the TEC was run in heating mode at steady state. The cold side temperature grew
to approximately 70 °C by the end of the test.
The TEC was run in heating mode at steady state with the cooling fan on. The
results of the test with a 10.0 V input are shown in Figure 2.7. The temperature of the
16
cool side was limited to 40 °C. The final temperature differential between hot and cold
side was 63 K as shown in Figure 2.7.
Figure 2.7 Steady state heating mode of the TEC with cooling fan and 10.0 V input
The TEC was next operated at steady state conditions in cooling mode with a 13.8
V input. The top of the TEC was the cold side during cooling while the hot side was
against the heat sink. The TEC was supplied with full power during this test and no
control system was used. The results were recorded until the system reached steady state.
Figure 2.8 Steady state cooling mode of the TEC without cooling fan and 13.8 V input
17
The steady state cooling mode of the TEC is shown in Figure 2.8. The final
temperature differential between hot and cold side was 42 K. It is noted that the cold side
trace in Figure 2.8 does not appear to be at steady state because this test was terminated
early. However, with the exception of an accurate measurement of the steady state
temperature difference, this test demonstrated the key aspects of interest.
A similar test with the cooling fan on is shown in Figure 2.9. The test shows the
cooling characteristics of the TEC from 94 °C. The test was run until steady state was
reached, as can be seen from the graph. The cold side of the TEC decreased to almost 0
°C. A temperature differential between top and bottom of approximately 50 K was
established.
Figure 2.9 Steady state cooling mode of the TEC with cooling fan and 13.8 V input
18
2.2.1 Heating flux characteristics of TEC
Modeling was done to better understand the heat flux characteristics of the TEC.
The data from the thermocouple placed on top of the TEC was used to calculate the heat
flux as a function of time during the heating and cooling modes of the TEC. The heating
of the top surface of the TEC which is made of aluminum oxide, also known as alumina,
is a transient problem, therefore transient methods must be used. The lumped
capacitance model is useful because it does not consider heat conduction within the solid.
The lumped capacitance method utilizes an energy balance around the solid. This
approach allows one to solve the transient problem easily. The lumped capacitance is
most accurate in situations where the Biot number is much less than one. If the Biot
number is much less than one, then convection from the surface occurs much more
readily than conduction within the solid, therefore conduction within the solid may be
ignored (Incropera and DeWitt 1996). Radiation effects were ignored in this case and the
convection coefficient was assumed to be constant with time. The lumped capacitance
method was selected to model this transient problem because the Biot number in this case
is 1.26 x 10-4, assuming that the convection coefficient is 5 W/m2K, the characteristic
length is 1 mm, and aluminum oxide has the properties shown in Table 2.1.
Table 2.1 Properties of aluminum oxide used in the analysis (Accuratus 2005)
Property Aluminum OxideThermal Conductivity (W/m K) 25Density (kg/m^3) 3720Specific Heat (J/kg K) 880
The use of the lumped capacitance method greatly simplifies the problem. For the
conditions of this problem, the lumped capacitance model has an analytical solution as
(Incropera and DeWitt 1996)
19
))exp(1(/)exp( atTTabat
TTTT
ii
−−−
+−=−−
∞∞
∞ (2.1)
where a and b are given by
VchA
a S
ρ= and
VcAq
b ss
ρ
"
= (2.2)
Equations 2.1 and 2.2 were solved implicitly for the heat flux using temperature
data from the test. This was possible because the temperature at the beginning and end of
each second was known and could be inserted into the equation for T and Ti. All of the
other variables are known, except for qs”. The properties used in the equation are the
alumina properties listed in Table 2.1. The other variables were
As=0.0009 m2 (0.030 x 0.030) m2, surface dimensions of alumina
V=9 x 10-7 m3 (0.030 x 0.030 x 0.001) m3, volume of alumina
The heat flux produced by the TEC during heating from 72 to 94 °C with cooling
fan as calculated with the lumped capacitance method is graphed in Figure 2.10 along
with the corresponding temperature of the top and bottom of the TEC. The TEC was on
at full power for the first five seconds of the cycle meaning that the duty cycle was 100%,
after that the duty cycle was controlled with the control system. The maximum heat flux
within the 72 to 95 °C range with a 13.8 V input and cooling fan on was 1.23 W/cm2.
The maximum heat flux occurred at t = 3 seconds. The heat flux decreased for the next
two seconds despite the fact that the TEC was on at full power.
20
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18Time (seconds)
q" (W
/cm
^2)
253035404550556065707580859095100
Tem
pera
ture
(°C
)
q"Top of TEC temperatureBottom of TEC temperature
Figure 2.10 Heat flux produced by thermoelectric cooler during heating from 72 to
94 °C and corresponding top and bottom of TEC temperature with 13.8 V input, cooling
fan.
The reason the heat flux increased to a maximum and then began to decrease,
despite the fact that the TEC was on at full power can be attributed to the characteristics
of the TEC. The relevant governing equations of a TEC are given by (Melcor 2005)
)2
(22
TGkG
IITNq Cc Δ−−=ρα (2.3)
)(2 TGINV Δ+= αρ (2.4)
ch qWq += & (2.5)
Where cq is the heat pumped at the cold surface of the TEC, hq is the heat pumped at the
hot surface of the TEC, N is the number of thermocouples in the TEC, α is the Seebeck
21
coefficient, I is the electrical current consumed by the TEC, Tc is the cold side
temperature, ρ is the resistivity of the bismuth telluride which makes up the TEC
junctions, G is the area divided by the length of a thermoelectric element, k is the thermal
conductivity of the bismuth telluride, ΔT is the difference in temperature between the hot
side and the cold side of the TEC, V is the voltage supplied to the TEC, and W& is the
power supplied to the TEC via the power supply.
These equations may be used to estimate the performance of the TEC and
determine trends in its performance. The calculated heating and cooling fluxes as well as
the current for the TEC were determined by inserting the recorded values for the hot and
cold side temperatures into equations 2.3-2.5 with the supply voltage value of 13.8 V.
The material properties of the bismuth telluride are variable with temperature and were
obtained from the manufacturer (Melcor 2005). The results of the analysis are shown in
Figure 2.11.
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (seconds)
Tem
pera
ture
(°C
)
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
q" (W
/cm
^2)
ThTcInput powerCooling fluxHeating flux
Figure 2.11 Calculated Performance of TEC using hot and cold side temperatures from
experiment
22
The hot and cold side temperatures are plotted on the left axis while the heating flux,
cooling flux, and input power per unit area are plotted against the right axis. The results
indicate that as the difference in temperature between the hot and cold side of the TEC
increases, the heat flux, input power, and cooling flux all decrease. The reason for this is
clear looking at Equations 2.3-2.5. The value of qc decreases as the temperature
difference increases because Tc decreases and ΔT increases. The power decreases also
because the current decreases as the difference in temperature between the hot and cold
side increases.
The reason that the heat flux increases and then decreases, even though the TEC
was operating at full power is because as the temperature difference between top and
bottom increases, the heating flux decreases as the above analysis shows. The initial
ramp of the heating flux may be attributed to transient operation of the device while
decreases of the heat flux after the first five seconds may be attributed to the fact that the
input power to the device was reduced by the control system.
The maximum heat flux calculated by the Melcor model was 3 times greater than
the actual heat flux. Melcor recommends that an insulating boundary be placed around
the TEC. The TEC was not insulated in this case leading to losses to convection. The
error could also be caused by the model not relating to transient conditions well. The
model was useful for looking at trends in the heating and cooling flux, however.
The heat flux information obtained from the above analysis was then applied to
the bottom surface of a system model using a two dimensional finite element heat transfer
program called Finite Element Heat Transfer (FEHT). The nodal temperatures on the top
surface of the model were then compared with the experimental data in order to verify
23
that the heat flux information was correct. The 2-D model geometry is shown in Figure
2.12. The alumina plate is 30 x 30 x 1 mm.
Figure 2.12 2-D geometry used for cooling and heating model. All dimensions in mm.
The results, superimposed on the graph for the experimental results, are shown in Figure
2.13. The calculated surface temperatures differ by at most 1.7 K from the measured
values. This result indicates that the lumped capacitance method is a valid method for
calculating the heat flux for transient heating.
253035404550556065707580859095
100
0 2 4 6 8 10 12 14 16 18
Time (seconds)
Tem
pera
ture
(°C
)
2-D modelTop of TEC
Figure 2.13 Calculated temperature for top of the TEC superimposed on graph of actual
temperature for top of TEC for 13.8 V input and cooling fan on.
Time dependent heat flux BC
Aluminum oxide
Three other outer surfaces convective heat transfer coefficient of 5 W/m2 K
24
A similar calculation was performed to analyze the heating characteristics of the
TEC with the heat sink cooling fan off and a 13.8 V input. The calculated heat flux is
shown in Figure 2.14. The maximum heat flux for the TEC during heating from 72 to
94 °C was 1.92 W/cm2 with the heat sink fan off and a 13.8 V input.
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18Time (seconds)
q" (W
/cm
^2)
253035404550556065707580859095100
Tem
pera
ture
(°C
)
q"Top of TEC temperatureBottom of TEC temperature
Figure 2.14 Heat flux produced by thermoelectric cooler during heating from 72 to
94 °C with 13.8 V input, cooling fan off
The heat flux for the case without the cooling fan on and a 13.8 V input reacts
differently then the case where the cooling fan was on. The heat flux for the case without
cooling fan increases for the first three seconds. The TEC was being operated with a
100 % duty cycle during this time. The TEC was pulsed after the three second point.
This differs from the case with the cooling fan on, where the heat flux increased and then
decreased, despite the fact that the TEC was on at full power. The temperature of the
25
heat sink in the case with no cooling fan was 30 K warmer than the case with heat sink
fan on. The fact that the heat sink was warmer and the temperature difference between
top and bottom of TEC was less, created a higher heat flux and made it possible for the
heat flux to not decrease when on at full power as explained in the discussion of Figure
2.10.
The heat flux calculated in Figure 2.14 was input into the 2-D model and the
simulation was run to predict temperature. The calculated values are superimposed on
the temperature data in Figure 2.15. The maximum error between the model and actual
temperatures was 1.8 K.
253035404550556065707580859095
100
0 2 4 6 8 10 12 14 16 18Time (seconds)
Tem
pera
ture
(°C
)
2-D model
Top of TEC
Figure 2.15 Calculated temperature from 2-D model for top of the TEC superimposed
on graph of actual temperature for top of TEC for 13.8 V input and cooling fan off.
26
2.2.2 Cooling flux characteristics of TEC
The cooling flux of the TEC was analyzed with the same method as used to
determine the heating flux of the TEC. The 2-D model and the properties for the material
were the same. The first case analyzed was the case with the cooling fan. The results
from the test are shown in Figure 2.16.
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
2.2
2.4
0 1 2 3 4 5 6 7 8 9 10
Time (seconds)
qc"
(W/c
m^2
)
253035404550556065707580859095100
Tem
pera
ture
(°C
)
qc"Top of TEC temperatureBottom of TEC temperature
Figure 2.16 Cooling flux produced by thermoelectric cooler during cooling from 94 to
54 °C, 13.8 V Input, cooling fan on
Figure 2.16 shows the cooling flux characteristics of the TEC as calculated with
the lumped capacitance method. The maximum cooling flux produced by the TEC was
2.1 W/cm2 for the case with cooling fan on and a 13.8 V input. The TEC was on at full
power through t= 6 s. The cooling flux increases, reaches a peak, and then decreases as
the temperature difference between the top and bottom of the TEC decreases. This is the
27
exact opposite behavior observed in the case where the TEC was in heating mode. The
cooling flux increased as the difference between the top and bottom of the TEC decreased
in the heating mode case. The equations governing the behavior of the TEC may once
again be used to understand the operation of the TEC. In the heating mode case, the
value of ΔT increases because ΔT = Th-Tc. The value of ΔT is increasing in the cooling
mode case also because the difference in temperatures at t = 0 seconds is a negative
number. As time proceeds, the value of ΔT becomes less negative, so it is indeed
increasing. Thus, what appears to be a contradiction between the heating mode and the
cooling mode case is actually the same case, so it makes sense that the cooling flux
decreases as the hot and cold side temperatures get close to one another.
The tabulated cooling flux results were next put into the 2-D model as a time
dependent heat flux on the bottom surface of the alumina. The results of that analysis are
shown in Figure 2.17.
253035404550556065707580859095
100
0 1 2 3 4 5 6 7 8 9 10Time (seconds)
Tem
pera
ture
(°C
)
Top of TEC 2-D model
Figure 2.17 2-D calculated temperature for top of the TEC superimposed on graph of
actual temperature for top of TEC for 13.8 V input and cooling fan on
28
The maximum error between the measured values of temperature and the
calculated temperature was 2.2 K, indicating that the lumped capacitance model is an
acceptable model for this system.
A similar procedure was used to calculate the cooling flux produced by the TEC
during cooling with the cooling fan off. The results are shown in Figure 2.18.
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
2.2
2.4
0 1 2 3 4 5 6 7 8 9 10
Time (seconds)
qc"
(W/c
m^2
)
253035404550556065707580859095100
Tem
pera
ture
(°C
)
qc"Top of TEC temperatureBottom of TEC temperature
Figure 2.18 Cooling flux produced by TEC during cooling from 94 to 54 °C, 13.8 V
input, cooling fan off
The maximum cooling flux for this period was 1.7 W/cm2. The cooling flux in
the case with the cooling fan behaves in a similar manner to the case with cooling fan on.
The maximum cooling flux is 0.4 W/cm2 lower in the case with no cooling fan, however.
29
The reason for this is because the heat sink is initially 18 K higher than the cooling fan
case meaning that ΔT is greater for the no cooling fan case.
2.2.3 Power consumption
The voltage and current consumed by the TEC were monitored via the display on
the power supply and a multimeter placed in series with the TEC.
The test used to determine the power and average power consisted of the TEC
mounted on the heat sink with the cooling fan on in one case and off in the other case.
Thermal grease was used between the TEC and heat sink. Control was provided by a
thermocouple placed on top of the TEC. The input voltage to the TEC for this test was
13.8 V. As the TEC was heated, the current was monitored on the power supply display.
The peak power for the TEC with and without cooling fan on was 23.7 W. The
average power over cycle for the TEC with cooling fan was 11.3 W. The average power
over cycle for the TEC without cooling fan was 7.9 W.
2.3 Independent TEC testing discussion
The results of the testing of the TEC are summarized in Table 2.2.
Table 2.2 Comparison of TEC with and without cooling fan
Max Heating
rate (K/s)
Average heating
rate (K/s)
Max Cooling
rate (K/s)
Average cooling
rate (K/s)
Max heating
flux (W/cm^2)
Max cooling
flux (W/cm^2)
Max Power consumption
(W)
Average power
consumption over cycle
(W)TEC with cooling fan 5.2 1.8 6.6 5.8 1.2 2.1 23.7 11.3TEC without cooling fan 5.8 2.0 5.2 4.3 1.9 1.7 23.7 7.9
30
The cooling fan produced a difference in the heating and cooling rates of the
TEC. The maximum and average heating rate were greater for the case without the
cooling fan. The reason for this is that the difference between the hot and cold side
temperatures of the TEC was less for the case without the heat sink fan on. This allowed
the top of the TEC to heat up more rapidly. The average and maximum cooling rates for
the TEC with cooling fan on were 1.5 K greater. This occurred because the difference
between hot and cold side temperatures was less for the case with cooling fan on.
The differences in heating and cooling rates translate into time for each cycle of
PCR. Using the TEC without the cooling fan would save 2.2 seconds in heating time for
each PCR cycle. Multiplying this value by 35 gives a total time saving of 77 seconds.
Using the TEC with cooling fan would save 2.5 seconds for each PCR cycle.
Multiplying this by 35 cycles gives a total time saving of 88 seconds for an entire PCR
run with 35 cycles. There is little benefit in using the cooling fan considering the time
savings.
The real savings in not using the cooling fan comes in power consumption. The
average power over cycle for the case without cooling fan was 3 W less than the case
with cooling fan on. The reason for this is because the average temperature of the TEC
remained warmer during operation without cooling fan. This made it possible to heat the
top surface of the TEC using less power. More power was required for cooling, but
overall the case with no cooling fan used less power.
A disadvantage of not using the fan on the heat sink was that the heat sink got
progressively warmer with each PCR cycle. This made the system slightly more difficult
to control.
31
The cooling fan on the heat sink produced an effect on the performance of the
TEC at steady state. The bottom of the TEC remained 29.4 K cooler in the case with the
heat sink fan on. The temperature differential between the hot and cold side of the TEC
with cooling fan was 17 K more than the TEC with no cooling fan. These results indicate
that the cool side of the TEC will rise above ambient during steady state operation. The
cooling fan keeps the bottom of the TEC cooler because it increases the convection
coefficient between the heat sink and the air.
2.4 Testing of a commercially available Peltier thermocycler, apparatus and
procedures
A commercially available thermocycler was tested to determine the characteristics
of the device. The thermocycler was a Techne Techgene thermocycler, model
FTGene5D rated at 120 or 230 V and 250 W shown in Figure 2.19.
Figure 2.19 Techne thermocycler analyzed in this section
32
The thermocycler was cooled and heated via 4 Marlow Industries Inc. TECs, part
number SP1975. On the bottom surface of the TECs was a 10.1 X 9.6 cm aluminum heat
sink with a fin height of 4.5 cm. On the top surface of the TEC an aluminum heating
block weighing 40.7 g was mounted. The dimensions of the heating block were 5.7 x 5.4
x 0.3 cm. The heating block had wells protruding from the surface. The wells were 11.9
mm high, had a diameter of 7.6 mm, and had a wall thickness of 0.8 mm. The heating
block contained 20 wells making it possible to do 20 different PCR reactions in one run
as shown if Figure 2.20.
Figure 2.20 Aluminum heating block mounted on top of TECs in Techne thermocycler
The thermocycler was tested using T type thermocouples. 10 µL of water was
inserted into a 200 µL tube. A thermocouple was then inserted into the solution and the
lid of the PCR tube was closed. This assembly was then inserted into one of the wells
located in the heating block. The thermocouple inside the tube measured the response of
33
the reaction vessel to the heating of the block. Another empty 200 µL tube was used to
measure the temperature of the heating block. A thermocouple was wedged between the
empty tube and the well in order to determine the temperature of the block. The lid of the
thermocycler was closed during the experiment.
The power to the device was measured by severing the power cable and inserting
a multimeter in series with the thermocycler to measure the current. The input voltage to
the device was 120 VAC.
2.4.1 Commercial Peltier thermocycler testing results
The commercial Peltier thermocycler was tested for heating and cooling rate. The
results of that test are shown in Figure 2.21.
Figure 2.21 Simulated PCR cycle using commercial Peltier thermocycler
45
50
55
60
65
70
75
80
85
90
95
100
225 240 255 270 285 300 315 330 345 360 375 390 405 420 435 450 465Time (seconds)
Tem
pera
ture
(°C
)
Inside 200µL tube 10 µLwaterBlock temperature,measured in well
34
The setpoint temperatures for this test were 94, 54, and 72 °C. The temperatures that
were measured via the thermocouples were all 2 K higher than these values. The error in
measurement is likely in the thermal cycler. The thermocouple is inserted into a hole in
the aluminum block and then glued into place. There may have been some air gaps or
other resistance present in this setup. The maximum heating rate within the well of the
heating block for the commercial device was 3.6 K/s. The maximum cooling rate turned
out to be 2.0 K/s. The average heating rate for the block of this device during heating
between setpoints 72 and 94 °C was 2.8 K/s and 1.8 K/s for heating between setpoints 54
and 72 °C/s. The average cooling rate for this device was 1.4 K/s.
2.4.2 Power consumption of the commercial device
The commercial Peltier unit was tested to determine the power requirements of
the device. The unit operated on 120 VAC. The current consumed by the device varied
during the different stages of operation. The current supplied to the device during
heating from 72 to 94 °C varied from 2.2 to 1.2 A with the largest current consumption at
the beginning of heating and the smallest current as the unit reached the setpoint. This
makes sense because it is expected that the control system of the device would scale the
power down as the setpoint temperature was reached. Current during heating between 54
and 72 °C peaked at 2.0 A during the initial phase of heating and decreased to 0.3 A as
the device neared the setpoint. During cooling, the current in the device ranged from 1.8
A to1.1 A as the setpoint was reached. The current consumed by the device at the 94 °C
plateau was 0.4 A, 0.3 A at the 72 °C setpoint, and 0.3 A at the 54 °C setpoint.
35
Considering these values gives a maximum power requirement of 264 W with an average
power over cycle of 145.0 W.
2.4.3 PCR results with the commercial device
PCR was run in the commercial device many times. The results presented in this
section are just a representative sample of those efforts. An experiment was run that
demonstrated the effect of elongation time on PCR results. As was discussed in Section
1.2, two different sets of primers were used. One primer set was for amplifying a piece
of DNA 879 bp long while the other primer set was for a 216 bp amplicon. The
elongation time was adjusted until the smaller section of DNA was amplified, but the
longer section was not. The protocol for these experiments is shown in Table 2.4.
Table 2.4 Protocol for elongation time experiment
879 bp amplicon reaction mixture (µL)
216 bp amplicon reaction mixture (µL)
Buffer 3.0 3.0MgCl2 1.0 1.0dNTP 4.5 4.5Primers 1.5 1.0DNA 1.0 1.0Taq Polymerase 0.3 0.3Albumine 3.0 3.0Water 10.7 11.2
Total 25.0 25.0
PCR was run in either 500 or 200 µL polypropylene PCR tubes with the heated lid on the
Techne thermocycler on to prevent solution evaporation. The PCR cycle for the first case
consisted of a 95 °C activation period for 15 minutes followed by 35 cycles of 30 seconds
at 94 °C, 30 seconds at 54 °C, and 40 seconds at 72 °C. A 15 minute final elongation
step at 72 °C completed the PCR run. The Techne control system only counts the plateau
36
time in the cycle time. For example, 30 seconds at 72 °C means that the thermocycler
plateaued at that temperature for 30 seconds. The time required to heat the block from 54
to 72 °C is not included. The results of this test are shown in Figure 2.22.
Lane Description0 DNA ladder1 500 µL tube, 10 µL solution, 879 bp amplicon2 500 µL tube, 10 µL solution, 216 bp amplicon3 200 µL tube, 5 µL solution, 879 bp amplicon4 200 µL tube, 5 µL solution, 216 bp amplicon5 200 µL tube, 5 µL of solution, 879 bp amplicon6 200 µL tube, 5 µL of solution, 216 bp amplicon
Figure 2.22 PCR reaction results for extension time test, 40 s elongation time
The results indicate that 40 seconds at 72 °C was sufficient to elongate both the 216 and
879 bp DNA fragment. The DNA ladder may be used to determine the length of the PCR
amplicon. The DNA ladder is a 1000 bp ladder in 100 bp increments. This means that
the line in lane 0 closest to the bottom of the picture is 25 bp, the next line is 100 bp, the
next line is 200 bp, the next line is 300 bp, all the way up to the first line that is the 1000
bp marker.
25 bp
100 bp
200 bp
1000 bp
37
The next test run in this sequence of tests to determine the amplification rate of
these two amplicons was to decrease the elongation time to 30 seconds, decrease the
denaturation time to 20 seconds, and decrease the annealing time to 20 seconds. The
results of that test are shown in Figure 2.23.
Lane Description0 DNA ladder1 500 µL tube, 10 µL solution, 879 bp amplicon2 500 µL tube, 10 µL solution, 216 bp amplicon3 200 µL tube, 5 µL solution, 879 bp amplicon4 200 µL tube, 5 µL solution, 216 bp amplicon5 200 µL tube, 5 µL of solution, 879 bp amplicon6 200 µL tube, 5 µL of solution, 216 bp amplicon
Figure 2.23 PCR reaction results for extension time test, 30 s elongation time
The results indicate that decreasing the elongation time to 30 seconds has no effect on
either amplicon. It is expected that as the elongation time continues to be decreased, the
longer amplicon will fail to be copied. Decreasing the denaturation and elongation times
from 30 to 20 seconds also had no effect on the reaction as a signal was received for each
tube.
25 bp
100 bp
200 bp
1000 bp
38
The third step in this test consisted of decreasing the denaturation and annealing
times to 15 seconds and decreasing the elongation time to 25 seconds. The results of that
test are shown in Figure 2.24.
Lane Description0 DNA ladder1 500 µL tube, 10 µL solution, 879 bp amplicon2 500 µL tube, 10 µL solution, 216 bp amplicon3 200 µL tube, 5 µL solution, 879 bp amplicon4 200 µL tube, 5 µL solution, 216 bp amplicon5 200 µL tube, 5 µL of solution, 879 bp amplicon6 200 µL tube, 5 µL of solution, 216 bp amplicon
Figure 2.24 PCR reaction results for extension time test, 25 s elongation time
The lack of a clear signal for any of the 879 bp amplicons in Figure 2.24 indicates that
the 25 second elongation time was a problem for the longer amplicon. Theoretically, the
copying rate for DNA polymerase is 50 bp/s, which would mean that it would be possible
to copy a 1250 bp segment in this time. The experiment that was run in this case
indicates that the rate of copy is around 29 bp/s.
An experiment was run to determine if the results shown in Figure 2.24 were due
to the decrease in annealing and denaturation times, or by the decrease in extension time.
25 bp
100 bp
200 bp
1000 bp
39
The denaturation and annealing time in this step remained at 15 seconds while the
elongation time was increased to 30 seconds, 5 seconds greater than the previous test.
The results of the test are shown in Figure 2.25.
Lane Description0 DNA ladder1 500 µL tube, 10 µL solution, 879 bp amplicon2 500 µL tube, 10 µL solution, 216 bp amplicon3 200 µL tube, 5 µL solution, 879 bp amplicon4 200 µL tube, 5 µL solution, 216 bp amplicon5 200 µL tube, 5 µL of solution, 879 bp amplicon6 200 µL tube, 5 µL of solution, 216 bp amplicon
Figure 2.25 PCR reaction results for extension time test, 30 s elongation time with 15 s
at denaturation and annealing steps
The results indicate that with an elongation time 5 s longer, the longer amplicon was
copied, even with the shorter denaturation and annealing steps. This result shows that the
limiting factor in PCR is the elongation phase. The copying speed of DNA polymerase is
a limiting factor in PCR speed.
Another method of visualizing PCR reaction success is by running the reaction
with ethidium bromide in the solution at a volumetric concentration of 1:4000 in a glass
25 bp
100 bp
200 bp
1000 bp
40
capillary. This eliminates the need to use gel electrophoresis to determine if the reaction
occurred. If the reaction occurred, the solution will fluoresce when placed in a UV light
enclosure. If it did not occur, then no fluorescence will be seen. Ethidium bromide is a
substance that fluoresces when exposed to UV light. Ethidium bromide has the ability to
attach itself to the DNA molecule (D'Amico, Paiotta et al. 2002). The ethidium bromide
undergoes an increase in flourescense when bound to the DNA molecule (Pyun and Park
1985). The concentration of ethidium bromide must be just right. If a large
concentration of ethidium bromide is used, then it will be impossible to distinguish
between a successful reaction and an unsuccessful reaction because ethidium bromide
fluoresces under UV light by itself. If too small an amount is used then no fluorescence
will be seen if the reaction occurred or if if did not. A picture of three capillaries in the
UV light hood is shown in Figure 2.26.
Figure 2.26 Glass capillaries run with a 1/4000 volumetric concentration of ethidium
bromide. 1) PCR solution before amplification, 2) PCR solution after amplification in
commercial thermocycler, 3,4) Unsuccessful reactions on microheater
41
The fluorescence is easily detectable in the successful PCR reaction, labeled 2 in Figure
2.26. There is no fluorescence in the reaction mixture before PCR, labeled 1, and the two
unsuccessful reactions run on the microheater, labeled 3 and 4 in Figure 2.26.
42
3. TESTING OF A CONVECTIVE THERMOCYCLER FOR PCR
A convective thermocycler was built in the lab to test the possibility of using a
device that directs an air flow past a sample to perform the heating and cooling for the
PCR reaction. The heating and cooling rate, heating and cooling flux, and power
consumption were monitored.
3.1 Testing of a convective thermocycler, apparatus and procedures
Two different configurations of the convective thermocycler were built. The first
variation is shown in Figures 3.1 and 3.2.
Figure 3.1 Convective thermocycler setup #1
17.1 cm OD AC fan
10.2 cm OD X 58.4 cm aluminum vent pipe
Copper-constantan type T thermocouple
InstruNet model 100 analog/digital input/output system
Omega engineering solid state relay part No. SSR240DC215
Transformer Type 3PN1010
43
Figure 3.2 Heater mounted within vent pipe
The second configuration was designed with two fans. One of the fans was used
for flowing air over the heating element while the other fan was used for cooling. The
purpose of this was to speed cooling. This double fan convective thermocycler is shown
in Figures 3.3 and 3.4.
Figure 3.3 Double fan convective thermocycler setup
Heating element removed from hair dryer
10.2 cm OD x 55.2 cm aluminum vent pipe
17.1 cm OD AC fan
Copper-constantan type T thermocouple
InstruNet model 100 analog/digital input/output system
Omega engineering solid state relay part No. SSR240DC25
Transformer Type 3PN1010
44
Figure 3.4 Heater element within double fan convective thermocycler
There are several similarities between the two convective thermocycler designs.
The convective thermocyclers described in this section were controlled with pulse width
modulation (PWM), similar to the thermoelectric cooler (TEC) described in Chapter 2. A
Visual Basic program was written to control the system. The software used a
proportional-integral-derivative (PID) controller scheme to control the fractional on-time
of the heater during each control period (1 sec control period used in this work). The
pipe centerline air temperature near the outlet was measured and was compared against
the setpoint to act as the controller input signal.
The fan on the single fan convective thermocycler was used to drive air through
the device. The air is heated as it flows past the heating element. The fan speed in the
single fan unit was also controlled with PWM in heating mode. The fan power was
turned on only 50% of the time to yield a reduced fan speed. During cooling, the heater
Ambient air channel
Thermocouple for temperature control
Heating element from hair dryer
Heated air channel
45
was turned off and the fan was run at full power to introduce the maximum flow of
cooling air.
The double fan convective thermocycler operated in much the same way as the
single fan convective thermocycler. The heater was controlled in an identical manner.
The difference between the two designs was that the double fan convective thermocycler
used two fans. The first fan was run at reduced speed as described above for the single
fan unit and was responsible for driving ambient air over the heating element to create a
stream of hot air in the main channel (second fan was off during heating). The second
fan was responsible for providing the cooling air flow. It was run at full speed during
cooling mode (the heating fan was turned off during cooling).
The software recorded data at a frequency of 1 Hz. The data recorded was the
pulse width, setpoint, and temperature at the outlet of the heater. An Extech Instruments
multimeter, model 22-816 was used to measure the voltage and current output of the
variable transformer.
The thermocouple for measuring the temperature of the air in the thermocycler
was placed along the axis of the pipe at a point 7.6 cm upstream from the end of the pipe.
3.2 Convective thermocycler test results
The maximum heating rate for the single fan convective thermocycler was 2.0
K/s. The maximum cooling rate was 1.8 K/s. A single simulated PCR cycle using the
single fan convective thermocycler is shown in Figure 3.5.
46
45
50
55
60
65
70
75
80
85
90
95
100
1325 1340 1355 1370 1385 1400 1415 1430 1445 1460 1475 1490 1505 1520 1535 1550
Time (Seconds)
Tem
pera
ture
(°C
)
Air TemperatureSetpoint
Figure 3.5 Simulated PCR cycle using the single fan convective thermocycler
The average heating and cooling rates were also calculated. The average heating
rate for heating between 54 and 72 °C for this setup was 1.2 K/s. The average heating
rate between 72 and 94 °C was 0.9 K/s. The average cooling rate for the single fan
convective thermocycler was 1.2 K/s.
The double fan convective thermocycler was tested to determine the heating and
cooling rates. The results of the test are shown in Figure 3.6.
Figure 3.6 Simulated PCR cycle using the double fan convective thermocycler
47
Vent pipe Mixing walls
Fan
Heater
Figure 3.7 Cross section of
single fan convective
thermocycler showing mixing
walls located within vent pipe
The maximum heating rate of the double fan convective thermocycler was 4.2 K/s. The
maximum cooling rate for this setup was 7.9 K/s. The average heating rates were 2.6 K/s
for heating between 54 and 72 °C and 2.0 K/s for heating between 72 and 94 °C. The
average cooling rate was 5.3 K/s. The heating rates are higher because the relative flow
rate and power input were set to obtain higher rates for the double fan unit. Cooling rates
are higher because the cooling fan flow rate is higher due to a more open flow path
obtained when the heater element is not present in the cooling flow path. These
characteristics were the motivation in designing the double fan unit.
One of the important characteristics of a convective thermocycler is temperature
uniformity across the area where the reaction will be occurring. The temperature should
be as uniform as possible to ensure a quality PCR reaction. The single fan convective
thermocycler was tested for temperature uniformity. Different ideas were tried in order
to make the radial temperature profile as even as possible. 2 in. thick foil and fiberglass
duct insulation was used in one test to cover the outside of the duct. This was done in
order to reduce the amount of heat loss
through the walls of the duct, thus
making the wall temperature closer
to the duct midpoint temperature.
Another variant of the convective thermocycler
was to place mixing walls within the duct in an
attempt to make the flow within the tube more
turbulent, thereby enhancing mixing as shown in
Figure 3.7.
48
Another tactic that was tested to make the temperature profile more uniform was to add
an additional vent pipe 61 cm long to the end of the convective thermocycler. The result
of the testing is shown in Figure 3.8.
Figure 3.8 Radial temperature measurements for single fan convective thermocycler
with heater and fan at duty cyle = 1
The configuration that provided the best results in terms of temperature uniformity was
the case where the pipe was covered with insulation, mixing walls were inserted into the
pipe, and no extension was used. The standard deviation for this case was 3.5 K,
compared to 3.6 K for the case with insulation, no mixing walls, and no extension. The
worst case scenario was the case with the extension, no insulation, and no walls. This
case had a standard deviation of 9.3 K. The effect of the mixing walls on the temperature
49
uniformity was very small, as can be seen by comparing the standard deviations for the
top two traces in Figure 3.8. Only 0.1 K separated the standard deviations of these two
cases. There was a large drop off in the standard deviation in temperature profile for the
case with mixing walls, no extension or insulation. The standard deviation for that case
was 5.1 K, indicating that the insulation was more important than mixing walls in
maintaining temperature uniformity.
3.2.1 Power consumption of convective thermocyclers
For the single fan unit, the power input to the fan was 108 VAC and 0.6 A. The
power supplied to the heater element was 78.4 VAC at 7.7 A. For the double fan
convective thermocycler, the heater was supplied with of 118.2 VAC which drew a
current of 11.6 A. The cooling and heating fans were supplied with 82.3 VAC each of
which drew a current of 0.4 A.
The power consumption of the single fan convection cycle was found by
multiplying the current, voltage and duty cycle. The power required to run the fan was
not taken into account for this calculation because the power required by the fan is small
compared to the power required by the heater. The power required to run the fan at full
power is only 10% of the power required to run the heater at full power for the single fan
convective thermocycler. The single fan convective thermocycler heater required 604 W
to operate when it was not pulsed. The average power over a cycle shown in Figure 3.5
was 255 W.
The power consumption for the double fan convective thermocycler was found
similarly. The power required for the heating and cooling fans was not taken into
50
account for this calculation. The power required to run the fan at full power is only 2.4%
of the total power required to run the heater of the double fan convective thermocycler at
full power. The double fan convective thermocycler required 1371 W of power when not
pulsed. The average power over cycle shown in Figure 3.6 was 645 W.
It is noted that the power requirements of these convective thermocylers is quite
high due to the fact that a large volumetric flow rate of heated air is dumped to the room.
Air heating/cooling requires significant air velocities to maintain high heat transfer
coefficients and such designs suffer from poor energy performance (high power
requirements) unless some recycling scheme is used to recycle the hot air instead of
throwing it away. In early designs, energy recycling was attempted but it was found that
such designs impacted the cycling speed. Air heating/cooling designs were found to
provide reasonably fast heating and cooing but at the cost of large power requirements.
3.2.2 Convection characteristics of the thermocycler
The velocity of the air within the two convective thermocyclers was found in
order to determine the heating power produced at the tube outlet. A control volume
analysis was used to determine the velocity of the air in the tube. A schematic of the
model used is shown in Figure 3.9.
Figure 3.9 Schematic of control volume used for air velocity determination
Fan
Heater q
AV Furnace duct
Control volume
Outlet
Inlet
51
where VA is the air velocity produced by the fan and q is the energy per unit time
transferred to the control volume by the heater. The control volume energy rate balance
reduces to (Moran and Shapiro 1999)
)(0.
oiCV hhmq −+−= (3.1)
assuming that the control volume is at steady state, ignoring kinetic and potential energy
effects, and ignoring heat transfer from the control volume. Here .
m is the mass flow rate
and is given by (Moran and Shapiro 1999)
νaveAV
m =.
(3.2)
where ih and oh are the enthalpies of the air at the inlet and outlet respectively. A is the
cross sectional area of the pipe, Vave is the average air velocity within the pipe, and ν is
the specific volume of the air in the mass flow rate equation. The specific volume of the
air may be determined by considering the air to be an ideal gas at atmospheric pressure.
The enthalpies were also found considering air to be an ideal gas.
Each convective thermocycler was run at steady state to determine the air velocity
during heating within the thermocycler tube. For these tests, the heaters on both the
single and double fan convective thermocycler were run at constant power (25% on
time). The heating fan on both units were run at the same settings that were used in the
thermal testing (reduced speed at 50 % power).
A test was run with both thermocyclers to determine the air velocity in cooling
mode. The fan on the single fan convective unit was run at full power with the heater run
at a 50 % duty cycle to determine the air velocity in cooling mode. The cooling fan of
the double fan convective thermocycler was also run at a full power during cooling. This
52
cooling fan did not have a heater obstructing its path, so the air velocity was determined
with the manufacturing specifications. The manufacturing specifications indicate that the
fan is capable of driving 235 cfm of air at 0 static pressure (Rotron 2005). This value
was used by considering an analysis of the flow in the vent tube. The pressure drop in
the vent tube may be found using the Moody friction factor equation (Incropera and
DeWitt 1996). The equation is
2/)/(
2aveV
DdxdPfρ
−= (3.3)
Where P is the pressure, x is the distance along the vent tube, D is the diameter of the
tube, ρ is the density of air taken at ambient, Vave is the average air velocity, and f is the
friction factor. The friction factor may be obtained from the Moody diagram and is
dependent on the Reynolds number. Some assumptions were made to get an idea of the
pressure drop in the tube. For an air velocity of 4 m/s, which is an overestimate
considering the specifications and the fan diameter, the Reynold’s number is 25,000.
This leads to a value for the pressure drop of 2.19 Pa using equation 3.3 and knowing the
diameter and length of the vent tube. The airflow is near the maximum value for the fan
considering the manufacturers static pressure versus airflow curve (Rotron 2005).
Therefore, the air velocity was found from the maximum volumetric flow rate by
dividing by the area of the fan.
The temperature at the outlet of the heating tube during steady state was recorded
in order to determine the enthalpy of the air at the exit during all air velocity
determinations except for the double fan convective thermocycler cooling fan. The inlet
temperature of the air was room temperature. This allowed the enthalpy to be set at the
inlet and the outlet. The control volume heat transfer was simply the power supplied to
53
the heater. The only unknown variable in Equation 3.1 was the air velocity. The results
of the air velocity analysis are shown in Table 3.1
Table 3.1 Results of air velocity analysis for single and double fan convective
thermocyclers
Single fan heater
Single fan cooler
Double fan heater
Double fan cooler
Air velocity (m/s) 0.5 0.7 1.2 3.4Fan input power (W) 32.4 64.8 16.5 32.9
The air velocities in Table 3.1 are also given with the fan input power in order to
demonstrate the relationship between the two. The fan input power was found by
multiplying the input voltage, current, and duty cycle of the fan. The air velocity of the
single fan cooler is lower than the double fan heater air velocity. The single fan cooler
fan was run at full speed while the double fan heater was run at 50 % duty cycle. The
flow path in the single fan convective thermocycler was more obstructed by the heater
than the double fan convective thermocycler because of the method of attachment. The
double fan heater also used 4 times less power than the single fan cooler. This
demonstrates that it is important to avoid obstructing the heating fan too much with the
heating coil. The double fan cooler supplied an air velocity that was nearly three times
greater than the double fan heater using only twice as much power. This was because the
double fan cooler had no obstruction in its path.
The next variable found to characterize the convection thermocycler system was
the Nusselt number. The reaction vessels that were placed inside the convection
thermocycler were cylindrical or cylinder like, therefore a cylindrical Nusselt correlation
was used. The correlation was (Incropera and DeWitt 1996)
54
4/1____
PrPrPrRe ⎟⎟
⎠
⎞⎜⎜⎝
⎛=
s
nmDD CNu (3.4)
where ____
DNu is the average Nusselt number over the surface of the cylindrical reaction
vessel, Pr is the Prandtl number, Prs is the Prandtl number evaluated at the surface
temperature of the vessel, and C, m, and n are constants based on ReD and Pr. The
temperature of the surface of the reaction vessel was unknown. This fact does not play
heavily in the analysis, however. The only parameter evaluated at the surface
temperature in Equation 3.4 is PrS. The largest temperature difference that could possibly
be encountered in this analysis is 40 K, the difference between the denaturation
temperature of 94 °C and the annealing temperature of 54 °C. The Prandtl number varies
by only 0.006 between these temperatures, therefore the surface temperature of the vessel
was assumed to be the setpoint temperature prior to the one being analyzed. For
example, if the air temperature within the tube was at the 72 °C setpoint, then the surface
temperature of the vessel was assumed to be 54 °C. All of the other parameters in the
equation were evaluated at the air temperature within the vent tube. The diameter of the
reaction tube in this case was taken to be 2.6 mm, which is the average diameter of the
bottom 2.5 mm of a 200 µL polypropylene PCR tube. The bottom of the tube was
measured only because this was the area that was filled when 10 µL of water were
inserted into it as shown in Figure 3.10.
55
Figure 3.10 200 µL PCR tube with 10 µL of water used in convective analysis. All
dimensions in mm
The Nusselt number and the average convection coefficient were found for the air
velocities shown in Table 3.1. The results are summarized in Table 3.2.
Table 3.2 Nusselt number and convection coefficients for single and double fan
convective thermocycler
3.2.3 System response with a test sample
The lumped capacitance method was used to derive an expression for the time
constant of a contained water sample in the convective thermocycler that is useful for
discussing the system dynamics. The lumped capacitance approximation is valid if the
Biot number is less than 0.1 (Incropera and DeWitt 1996). In this case, the Biot number
56
was evaluated for a cylindrical reaction vessel with an average diameter of the bottom 2.5
mm of a 200 µL polypropylene PCR tube. The polypropylene tube is 0.3 mm thick. The
thermal conductivity of this system was lumped together based on the volumetric
percentage of water and polypropylene. This gives a thermal conductivity of
0.47 W/m K, considering that the thermal conductivity of water is 0.60 W/m K and
polypropylene is 0.17 W/m K. Evaluating the Biot number with these constraints gives a
value of 0.17 with an average convection coefficient of 120 W/m2K and 0.1 with an
average convection coefficient of 71 W/m2K. These results indicate that the lumped
capacitance model may be a poor choice for the cooling case in the double fan convective
thermocycler, but may be a good fit for the other situations. Applying an energy balance
on the reaction tube and sample, the following equations can be derived (Incropera and
DeWitt 1996)
⎥⎦
⎤⎢⎣
⎡⎟⎟⎠
⎞⎜⎜⎝
⎛−=
−−
∞
∞ tVc
hATTTT s
i ρexp (3.5)
where T is the temperature at time t, Ti is the initial temperature of the system, ∞T is the
air temperature within the vent pipe, h is the convection coefficient, ρ is the density of the
reaction vessel, V is the volume of the reaction vessel, c is the specific heat of the
reaction vessel, and As is the surface area of the reaction vessel. This expression gives
rise to an expression for the time constant of the reaction vessel
ttt CR=τ (3.6)
The equation states that the time constant of the system is equal to the resistance to
convection multiplied by the total thermal capacitance of the system. The analysis that
follows will be performed first on a 200 µL polypropylene PCR tube with 10 µL of water
57
inside of it modeled as a cylinder of diameter 2.6 mm and height 2.5 mm. The time
constant in this case is given by
s
ppppwpwwt Ah
fcfcV )( ρρτ
+= (3.6)
where V is volume of the reaction tube, ρw and ρp are the densities of the water and
polypropylene, cpw and cpp are the specific heats of the water and polypropylene, As is the
surface area of the cylinder, and fw and fp are the fraction volume of water and
polypropylene in the reaction volume. A lumped thermal conductivity was found using
the volumetric fraction of water and polypropylene that was used to determine the
average heat transfer coefficient from DNu .
ppww fkfkk += (3.7)
Time constants were found for the system using the average convection
coefficients found in Table 3.2. The results are shown in Table 3.3.
Table 3.3 Calculated time constants (τ) for the air velocities within the single and double
fan convective thermocycler for a modeled 200 µL polypropylene tube with 10 µL of
water
The time constant is the time that it takes the reaction volume to reach within 63.2 % of
the difference between the initial temperature of the solution and the setpoint
temperature. If the setpoint was 72 °C, the initial temperature was 52 °C, and the time
constant was 34 seconds, the reaction should be at 64.6 °C after 34 seconds, for example.
58
A 200 µL polypropylene PCR tube filled with 10 µL of water was modeled using
a 2-D heat transfer finite element program. A schematic of the model used is shown in
Figure 3.11.
Figure 3.11 2-D model used for convective analysis of 200 µL polypropylene tube filled
with 10 µL of water, dimensions in mm
The temperature profiles of the convective thermocyclers obtained in the results section
were input into the 2-D model shown in Figure 3.11 along with the time average heat
transfer coefficients, calculated with the velocity and temperature information from
experiment. The experimental results were then compared with the 2-D model results to
determine the validity of this method. The properties of polypropylene used in the model
are shown in Table 3.4 (Goodfellow 2005).
Table 3.4 Polypropylene properties
Density (kg/m3)
Thermal conductivity
(W/m K)
Specific heat (J/K kg)
Polypropylene 900 0.17 1800
59
Heating from 52 to 94 °C was modeled with the 2-D finite element program for
the single fan convective thermocycler. The experiment with a 200 µL tube with 10 µL
of fluid in the single fan convective thermocycler indicated that the initial temperature of
the fluid before the heating began was 53.3 °C as shown in Figure 3.12.
Figure 3.12 Experimental results for single fan convective thermocycler with 200 µL
tube filled with 10 µL of water placed inside vent pipe
The temperature of the fluid inside the tube was measured with a thermocouple placed
inside the tube, immersed in the fluid. Analysis showed that the average convection
coefficient over the surface of the cylinder remained relatively constant during heating
from 52 to 94 °C, varying by only 0.06 %. A constant value of h , average convection
coefficient, was used for this reason and was taken to be 48.0 W/m2K.
The 2-D model produced the results shown in Figure 3.13.
60
Figure 3.13 2-D model correlation with experimental results for single fan convective
thermocycler with 200 µL PCR tube containing 10 µL of water during heating from 52 –
94 °C
The maximum error between the experimental value and the modeled value was 3.6 K.
The calculated time constant for the system was 34 seconds using the lumped capacitance
model. This means that the temperature at 34 seconds should be 64.6 °C for the heating
case between 52 and 72 °C. The actual water temperature was 65.2 °C at this point while
the 2-D finite element model predicted a temperature of 65.7 °C. The significant time lag
of the reaction mixture in comparison with the air temperature is shown in Figure 3.13.
The air temperature reached the setpoint of 72 °C within 8 seconds while the reaction
mixture failed to reach the setpoint temperature within the 60 second time frame.
The single fan convective thermocycler with 200 µL PCR tube filled with 10 µL
of water during cooling from 86 to 52 °C was modeled with the 2-D finite element
61
modeling package. The initial temperature of the solution at the beginning of the
temperature decrease was 86.2 °C. The convection coefficients used were 57 W/m2K for
the first fifteen seconds of the cool down, corresponding to the time when the fan was on
at full power, and 48 W/m2K for the period after 15 seconds when the fan was pulsed.
The results of the analysis are shown in Figure 3.14.
Figure 3.14 2-D model correlation with experimental results for single fan convective
thermocycler with 200 µL PCR tube during cooling from 86 – 52 °C
The calculated time constant for this reaction volume during cooling was 29
seconds, indicating that the solution temperature should be at 64.5 °C after one time
constant. The temperature of the solution was actually at 62.1 °C at this time. The 2-D
modeled temperature was at 71 °C after 29 seconds. The maximum error between the
model and the actual temperatures was 8.8 K, indicating poor agreement between the
62
model and experimental data. This result indicates that the actual air velocity in the tube
may be higher than calculated using the control volume approach.
The double fan convective thermocycler reaction volume was modeled with the
lumped capacitance method in order to validate this method for future work. The
performance of the double fan convective thermocycler with a 200 µL polypropylene
tube filled with 10 µL of water is shown in Figure 3.15.
Figure 3.15 Experimental results for double fan convective thermocycler with 200 µL
polypropylene tube filled with 10 µL of water placed inside vent pipe
Recall that the polypropylene tube was modeled as a cylinder with a diameter equal to the
average 2.5 mm of the reaction tube and a wall thickness of 0.3 mm. The time constant
for the air in the vent tube was 8 seconds for the heating modes and 6 seconds for the
cooling mode. The calculated time constants for the fluid in the reaction tube are 23
63
seconds for the heating mode and 14 seconds for the cooling mode when the fan is on at
full power as shown in Table 3.3.
The temperature at time t is dependent on ∞T , the temperature of the air stream, as
shown in Equation 3.5. The temperature of the air stream is transient during the initial
part of the heating phase. This gives two views on how to model this problem. The first
view is to model the problem with the transient air temperature of the air, substituting the
appropriate value of ∞T (t). The second view is to model the problem using constant
value representing the setpoint for ∞T since the time constant of the reaction volume is 3
times greater than the time constant of the air within the vent tube during heating. These
two approaches were used for the heating modes of the double fan convective
thermocycler. The results of the modeling are shown in Figure 3.16.
45
50
55
60
65
70
75
80
85
90
95
100
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150Time (seconds)
Tem
pera
ture
(°C
)
Air temperature
Water temperature
Lumped capacitancemodel, constant Tinf
Lumped capacitancemodel, transient Tinf
Figure 3.16 Lumped capacitance model with experimental results for double fan
convective thermocycler with 200 µL PCR tube during heating from 54–94 °C
64
The correlation between experimental data and the lumped capacitance models is
strong. The lumped capacitance model with constant ∞T has more error associated with
it then the model with transient ∞T during the first eight seconds of the test, and again
during the first eight seconds of heating from 72 to 94 °C. The average error over the
first 8 seconds of heating from 54 to 72 °C for the constant ∞T case was 1.0 °C while the
average error associated with the transient ∞T case was 0.4 °C. The average error over
the first 8 seconds of heating from 72 to 94 °C for the constant ∞T case was 2.0 °C while
the average error for the transient ∞T case was 0.5 °C. The average error for the
remainder of the time period was also calculated. The average error for the transient ∞T
case turned out to be 0.5 °C while the average error for the constant ∞T case turned out to
be 0.6 °C. The time constant of the reaction volume for heating between 54 and 72 °C
was 20 seconds. The time constant of the reaction volume for heating between 72 and 94
°C was 25 seconds. The calculated time constant was 23 seconds for these temperature
intervals. These results indicate that accurate results can be obtained by ignoring the
transient heating of the air in modeling the solution temperature with the lumped
capacitance method as long as the time frame being analyzed is outside the time constant
of the air, in this case 8 seconds, and the time constant of the reaction volume is 3 times
larger than the time constant of the air.
The lumped capacitance model was used to model cooling of the reaction volume
consisting of a 200 µL PCR tube containing 10 µL of water within the double fan
convective thermocycler. The same approach was used as was used for the heating
modes of the double fan convective thermocycler. The first situation was when ∞T was
65
held constant at the setpoint temperature. The second situation was when ∞T was allowed
to vary with the air stream temperature. The cooling phase consisted of two different
values of h . The first value was the one associated with the fan at full power and was
taken to be 120 W/m2 K. Recall that during cooling, the cooling fan is turned on at full
power and the heater fan is turned off. The second value of h was associated with the
condition when the cooling fan was turned off and the thermocycler returned to its
normal operating state, the heating fan pulsed with the heater on. The second value of h
was taken to be 71 W/m2K. The cooling fan was on at full power during the test for
seven seconds. The results of the modeling using the lumped capacitance method are
shown in Figure 3.17.
45
50
55
60
65
70
75
80
85
90
95
100
0 15 30 45 60 75Time (seconds)
Tem
pera
ture
(°C
)
Air Temperature
Water temperature
Lumped capacitance model,constant TinfLumped capacitance model,transient Tinf
Figure 3.17 Lumped capacitance model with experimental results for double fan
convective thermocycler with 200 µL PCR tube during cooling from 94–54 °C
66
The time constant for the cooling air was 6 seconds because it took the air
temperature that long to reach 69 °C. The time constant for the reaction volume was 17
seconds. The calculated time constants for the fan at full power and the heating fan at 0.5
duty cycle were 14 and 23 seconds, or an average of 18 seconds, which corresponds well
with the experimental results. It is clear by looking at Figure 3.17 that the error
associated with the constant ∞T case was much greater than the error associated with the
transient ∞T case over the six second time constant of the air temperature. The average
error for the time period minus the initial six seconds for the transient ∞T case was 2.4 °C
while the average error for the constant ∞T case was 1.4 °C. This indicates, as does the
heating case, that the transient cooling of the air may be ignored when modeling the
reaction volume using the lumped capacitance method for times much greater than the
time constant of the air. If the reaction volume temperature is desired during a time less
than the time constant of the air, than the transient ∞T lumped capacitance model should
be used. Both lumped capacitance models in Figure 3.17 have a small area where the
curve is not smooth at t = 8 s. This is because this is the point where the heat transfer
coefficient was changed, corresponding to the time when the cooling fan was turned off,
and the heating fan was turned on, pulsed at 0.5 duty cycle.
3.3 Convective thermocycler options to enhance performance
Increasing the heating rate of the air within the vent tube is one way to enhance
the performance of the thermocycler. Considering the heater itself, a heater with greater
surface area could be designed to allow for more efficient heat transfer from the heater to
67
the air. More power could also be supplied to the heater to add more energy to the
system.
High air velocity and high heating rate of the air is a challenge because the heater
can only supply so much energy to the air as it flows by it. Two important parameters for
high reaction volume heating rate are velocity and air temperature heating rate. The
velocity of the air may only be increased by so much before the heater will not be able to
heat the air to the setpoint temperature.
An increase in air velocity within the convective thermocyclers would increase
the Reynolds number and would therefore increase the convection coefficient between
the tube and the air. The time constant of the reaction was determined for a variety of
different air velocities assuming a glass cylinder with wall thickness of 240 µm and
outside diameter of 890 µm corresponding to the diameter of a commonly found glass
capillary, an air stream temperature of 94 °C, a surface temperature of 72 °C, and a length
of 2.6 mm. The capillary was filled with water. The plot of the time constant versus air
velocity is shown in Figure 3.18.
0
1
2
3
4
5
6
7
0 1 2 3 4 5 6 7 8 9 10Air velocity (m/s)
Tim
e co
nsta
nt (s
)
Figure 3.18 Plot of time constant versus air velocity for cylindrical configuration with
constant diameter, length, wall thickness, surface temperature, and air temperature
68
All of the time constants, except one, graphed above are lower than the heating time
constant of the convective thermocycler which was 8 seconds. The cooling time constant
of the double fan convective thermocycler as tested was 6 seconds. All of the graphed
time constants are less than this also, except for the 0.5 m/s case. The change in time
constant for this reaction vessel versus air velocity levels out somewhat after 4 m/s. The
benefit of increasing the air velocity after this point results in reduced return, compared
with the benefit obtained before the 4 m/s point.
The reaction volume may be manipulated to enhance the heat transfer. One way
to do this for a cylindrical reaction vessel is to decrease the diameter of the reaction
vessel. It was shown above that a cylindrical glass tube with an outside diameter of 890
µm has a shorter time constant compared with the 2.6 mm diameter polypropylene tube.
A plot of time constant versus cylinder diameter was made assuming a constant air
velocity of 1.2 m/s, a wall thickness of 244 µm, a polypropylene tube, and a length of 2.6
mm is shown in Figure 3.19.
0
50
100
150
200
250
300
350
0.0 5.0 10.0 15.0 20.0Cylinder diameter (mm)
Tim
e co
nsta
nt (s
)
Figure 3.19 Plot of time constant versus diameter for cylindrical polypropylene
configuration
69
This plot shows that the time constant with a cylinder of this diameter would be 5
seconds which is less than the double fan convection air temperature time constant of
eight seconds. The cylindrical tube studied in Secion 3.2.3 is the point with diameter =
2.62 mm corresponding to a time constant of 23 seconds. This analysis shows the
dramatic effect that changing the diameter of the reaction vessel can have on the time
constant of the system. The cost of decreasing the diameter of the reaction vessel is that
one can get less solution into it per unit length.
The lumped capacitance method was used to model the behavior of a glass
capillary in the double fan convective thermocycler. The inside diameter of a commonly
available glass capillary is 0.4 mm so it would be necessary to have a capillary 39.8 mm
long to have 5 µL of PCR solution in it. For a reaction volume of this geometry with this
amount of water in it, a wall thickness of 244 µm, and an air velocity of 1.2 m/s the
lumped capacitance method yields a time constant for the system of 4.8 seconds, a value
of 3.6 for the average Nusselt number, and 126 W/m2K for the average convection
coefficient. This time constant was used to model the behavior of the reaction volume in
the double fan convective thermocycler as tested previously. The results of the analysis
are shown in Figure 3.20.
70
Figure 3.20 Lumped capacitance model of reaction volume temperature for glass
capillary of 890 µm outside diameter with 5 µL of water inside
These results indicate that the solution temperature is nearly identical to the air
temperature for the case with the glass capillary in the double fan convection
thermocycler. This indicates that the limiting factor in PCR speed is the ability of the
thermocycler to heat the air since the time constant of the reaction mixture is less than
that of the air temperature within the thermocycler.
The reaction mixture may also be placed into a reaction vessel of different shape.
One reaction vessel that is of interest is a sphere. A sphere of inside diameter 2.12 mm is
needed to have a reaction volume of 5 µL. the average Nusselt number is given by
(Incropera and DeWitt 1996)
71
4/14.03/22/1 Pr)Re06.0Re4.0(2 ⎟⎟
⎠
⎞⎜⎜⎝
⎛++=
sDDDNu
μμ (3.8)
The lumped thermal conductivity was used in the calculation of the average heat transfer
coefficient as in Equation 3.7. A glass sphere with this inside diameter and wall
thickness of 244 µm, an air velocity of 1.2 m/s, a surface temperature of 72 °C, an air
stream temperature of 94 °C, and the lumped capacitance method, a time constant of 16.1
seconds is obtained with an average Nusselt number of 7.2, and an average value of the
convection coefficient of 90 W/m2K. The time constant for the spherical geometry is
greater than the 890 µm outside diameter cylinder with the same reaction volume because
the diameter of the cylinder is much less than the sphere. For the geometries used in this
analysis, the surface area to volume ratio of the cylinder is almost twice as great as the
sphere. Even though the average Nusselt number is twice as great for the sphere, the
convection coefficient convection is smaller because of the smaller surface area to
volume ratio of the sphere. The time constant of the spherical system versus diameter for
a 1.2 m/s air velocity, glass wall thickness of 244 µm, surface temperature of 72 °C, and
air stream temperature of 94 °C is plotted in Figure 3.21.
72
0
10
20
30
40
50
60
70
80
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0Sphere diameter (mm)
Tim
e co
nsta
nt (s
)
Figure 3.21 Plot of time constant versus diameter for spherical configuration with
constant air velocity, and wall thickness
The materials and dimensions of materials needed to optimize the reaction vessel
were found. The lumped capacitance model was used to determine all of the time
constants above. It is clear that the lumped capacitance model is an accurate model of
what is occurring as was seen in its comparison with actual experimental data. The
lumped capacitance model is an idealization that ignores conduction within the reaction
volume. This is the best case scenario because if the reaction volume was large enough
such that the lumped capacitance model was not valid, then the reaction dynamics would
be slowed by conduction within the volume. The lumped capacitance model is valid
under the following condition (Incropera and DeWitt 1996)
1.0<=k
hLBi c (3.9)
73
where Lc is the characteristic length and is given by the volume to surface area ratio of
the reaction volume. The variable k is taken to be the lumped thermal conductivity of the
reaction volume as in Equation 3.7.
The Biot number may be used to determine what reaction tube materials would
allow for quick heating. Assuming a convection coefficient of 126 W/m2K, which
corresponds to the convection coefficient for the double fan convective thermocycler
during heating for a cylindrical reaction vessel with 244 µm thick walls and an outside
diameter of 890 µm, gives a value of 0.20 W/m K as the minimum thermal conductivity
to satisfy the Biot number. Common materials used to build reaction vessels have
thermal conductivities at or above this value. The thermal conductivities of several
materials that could be used to make a reaction vessel are shown in Table 3.5 (Incropera
and DeWitt 1996; Efunda 2005).
Table 3.5 Thermal conductivities of candidate materials for PCR reaction vessels
Material Thermal Conductivity (W/mK) Polypropylene 0.17 Polycarbonate 0.20 Silicon 148 Glass (soda lime) 1.4
The thermal conductivities of the materials listed in Table 3.5 are all greater or very close
to 0.2 W/m K so any one of them could be used in the cylindrical configuration
mentioned above without causing large spatial temperature gradients within the reaction
vessel.
74
The diameter of the reaction vessel may also be varied. It is of interest to know
the largest diameter possible of the cylindrical reaction vessel such that spatial variation
in temperature does not play into the analysis. This calculation was run assuming that the
velocity of the air was 1.2 m/s, the cylinder was 39.8 mm long, the cylinder was being
heated from 72 to 94 °C, and the material was polypropylene with a wall thickness of 244
µm. The results of this analysis are shown in Table 3.6.
Table 3.6 Biot number for varying polypropylene cylindrical diameters
Cylinder Diameter (mm)
Average convection coefficient (W/m2K) k (W/mK) Bi
0.89 126. 0.26 0.11
1.0 118 0.28 0.10
2.0 82 0.42 0.10
3.0 66 0.47 0.11
4.0 56 0.50 0.12
5.0 50 0.52 0.12
6.0 45 0.54 0.13
7.0 42 0.55 0.13
The results indicate that the Biot number for the cylinders is right around 0.1. The Biot
number varies little for the diameters input above because as the convection coefficient
decreases, the value of k increases because of the greater percentage of water, and the
characteristic length increases, the net result is little change in the Biot number. These
results indicate that varying the diameter of the cylinder up to 7 mm under these
conditions has little effect on the Biot number, indicating that spatial temperature
75
gradients should be small. The problem with increasing the diameter of the tube is the
decreased convection coefficient.
The Biot number becomes larger than 0.1 if the wall thickness of the capillary is
increased. A polypropylene capillary with an air velocity of 1.2 m/s flowing past it
during heating from 72 to 94 °C, and a length of 40.0 mm was analyzed by varying the
tube wall thickness. The results of the analysis are shown in Figure 3.22.
Figure 3.22 Biot number versus wall thickness for a poly propylene capillary of outside
diameter 890 µm
Increasing the wall thickness of a polypropylene capillary under these conditions can
cause an increase in the Biot number, which is an indicator of the temperature gradient
that will be present within the reaction vessel. The reason for this is that the low thermal
conductivity of polypropylene begins to dominate the reaction dynamics because of its
thickness in relation to the water. If the capillary material is glass, which has a greater
76
thermal conductivity than water, than the Biot number decreases with increasing wall
thickness as shown in Figure 3.23.
0
0.005
0.01
0.015
0.02
0.025
0.03
0.035
0.04
240 250 260 270 280 290 300 310 320 330 340 350 360 370 380 390 400
Wall thickness (µm)
Bio
t num
ber
Figure 3.23 Biot number versus wall thickness for a glass capillary of outside diameter
890 µm
The decreasing Biot number means that the temperature of the reaction volume will be
nearly uniform during the thermocycling process. Using materials with a higher thermal
conductivity than water can be beneficial in making the reaction vessel more isothermal.
77
4. ANALYSIS OF MICROHEATERS FOR PCR
Another option for heating reaction volumes for PCR is the use of microheaters
(Selvaganapathy, Carlen et al. 2003; Lee, Ho Park et al. 2004). Microheaters are heaters
that have features that can be measured on the micron scale. Microsystems take
advantage of reaction volumes on the µL or in some cases nL scale to enhance the speed
of PCR (Matsubara, Kerman et al. 2004; Gulliksen, Solli et al. 2005). Reaction volumes
that have low thermal capacitance allow rapid PCR by minimizing energy storage within
the vessel. The microheater is typically tightly coupled to the reaction volume (Yoon,
Lee et al. 2002), allowing heat transfer to occur rapidly, unlike Peltier devices which are
typically covered with a ceramic layer, and in some cases have metal blocks mounted on
top of them.
The work of other groups in this area was studied and a survey of that work is
described in Section 4.1. Microheaters were designed and fabricated using standard
MEMS techniques, as discussed in Section 4.2 . A control system, discussed in Section
4.3, for temperature cycling with the microheater was built and tested.
4.1 Synopsis of work done on microheaters by other groups
SU-8 is one option for creating microchambers on top of microheaters. It has
been used as a mold for PDMS microchambers (Zhao, Cui et al. 2002). El-Ali and others
have recently incorporated a thin film platinum heater with an SU-8 reaction volume to
achieve heating and cooling rates five times faster than the fastest macroscale device and
power consumption 126 times less (El-Ali, Perch-Nielsen et al. 2004). The device
created by this group consists of an SU-8 based PCR chamber that is affixed to a 1 mm
78
thick glass substrate with a thin film platinum heater deposited on it. There are 83
individual heater elements, each element is 20 µm wide and 8.3 mm long with spacing
between heater elements of 100 µm. The heaters extend beyond the boundaries of the
reaction chamber to minimize cold wall effects. The glass also has an independent
thermometer separate from the heater itself. The resistance of the platinum thermometer
changes linearly with temperature making it possible to determine the temperature by
measuring the resistance. The reaction chamber is 7 x 7 x 0.4 mm, giving a total reaction
volume of 20 µL. Atop the reaction chamber a 0.5 mm glass piece is bonded with SU-8
utilizing SU-8 to SU-8 bonding. The lid has two small openings in it, through which the
reaction chamber may be filled. It was not mentioned how these holes were sealed
during cycling. The chip is cooled via natural convection and through a heat sink
connected to the lower surface. A Labview PID controller was used to control the
heating and cooling of the chip. A schematic of the device is shown in Figure 4.1 (El-
Ali, Perch-Nielsen et al. 2004).
Figure 4.1 Heater with integrated well as designed by El-Ali et al.
79
El-Ali et al. used standard MEMS techniques to fabricate the device shown above.
The heater electrodes and the thermometer were patterned with lift-off on the glass
substrate. The platinum was deposited via e-beam evaporation using a 100 Å titanium
layer for adhesion. A 5 µm SU-8 layer was spun onto the wafer. The pattern was soft
baked and patterned with photolithography. The purpose of this layer was to protect the
electrodes from the PCR solution. The 400 µm deep reaction vessel was defined with
two spins of 200 µm each. A soft bake step was used in between each spin. The glass lid
was next bonded to the SU-8 reaction vessel using SU-8 to SU-8 bonding by first
applying a thin coat of SU-8 to the glass lid, soft baking the lid with SU-8 on it,
positioning the lid on top of the SU-8 section, soft baking, exposing with UV light, and
finally another bake. It was necessary for the researchers to treat the SU-8 surfaces of the
reaction chamber with the silanizing agent dichlorodimethylsilane to make it PCR
compatible. The dichlorodimethylsilane makes the surface PCR compatible by inhibiting
adsorption of DNA and polymerase to the surface of the chamber (Kopp, de Mello et al.
1998). The final step in the production of the chip was drilling two holes in the glass lid
for filling of the reaction vessel (El-Ali, Perch-Nielsen et al. 2004).
Silicon has also been tested as a material for microchambers (Daniel, Iqbal et al.
1998; Nagai, Murakami et al. 2001; Matsubara, Kerman et al. 2004). Zhao et al.
designed a PCR microchip with an integrated heater utilizing silicon as the material for
the reaction vessel (Zhao, Cui et al. 2003). The device developed by this group was an 8
x 4 mm chip. The reaction vessel is 4 x 2 x 0.27 mm, giving a total reaction volume of 2
µL. The reaction vessel was manufactured on one side of the chip while the thin film
platinum heater was deposited on the opposite side of the chip. The heater was separated
80
Figure 4.2 PCR reaction vessel designed by Zhao et al. (Zhao, Cui et al. 2003).
from the reaction volume by a distance of only 2 µm. Two
channels were etched on opposite sides of the reaction
vessel to provide a method for filling the vessel with PCR
reagents. The chip was anodically bonded with a 0.2 mm
thick piece of borosilicate glass. Two holes were drilled in
the glass and were aligned with the chip to allow filling.
A schematic of the setup is shown in Figure 4.2. The
device consists of a heating power supply circuit,
temperature controller, keypad, and display. The heater was controlled with a
microcontroller and a PID scheme. The heater power was controlled with pulse width
modulation at a frequency of 19.5 kHz (Zhao, Cui et al. 2003).
The device was fabricated using a 270 µm double-side polished silicon wafer.
Prior to etching, silicon nitride was deposited on each side of the wafer to a thickness of 2
µm. The reaction vessel was wet etched with KOH utilizing the silicon nitride as a stop
layer. A 100 nm oxide layer was deposited after the wet etch to provide a suitable
surface for the PCR reaction. Two 1 mm length channels were etched on each side of the
reaction vessel to allow filling. The platinum heater and temperature sensor were
patterned on the backside of the wafer using standard photolithography and lift off. The
chip was next mounted on a PCB board with patterned electrodes to connect to the heater
(Zhao, Cui et al. 2003).
Researchers have turned to plastic microchambers for use with microheaters
(Giordano, Ferrance et al. 2001; Liu, Rauch et al. 2002). Zou et al. have taken this
approach to designing a PCR microchip with integrated heater. The device consists of
81
multiple silicon blocks with a thickness of 680 µm bonded to a 500 µm PCB substrate
with a heat sink attached to the bottom. Heaters and temperature sensors are fabricated
on the underside of the silicon blocks. The reaction chambers were thermally molded in
plastic with a silicon plate etched with KOH. The type of plastic used was not specified.
The 4 x 4 array of reaction chambers are affixed to the top of the heaters with thermal
grease and mechanical means. The reaction chambers are 400 µm deep, are
approximately 5 mm square, and have a wall thickness of 100 µm yielding a reaction
volume of 20 µL. A schematic of the device is shown in Figure 4.3 (Zou, Miao et al.
2002).
Figure 4.3 Microheaters with integrated reaction vessels as designed by Zou et al.
The heater array manufactured by Zou used standard MEMS techniques. The
process began with a thermal oxide deposition on the top of the wafer. The heaters and
temperature sensors were manufactured by depositing the aluminum to a thickness of 2.4
µm and then patterning. Plasma enhanced chemical vapor deposition was next used to
deposit a silicon nitride layer on top of the heater elements. This layer was patterned to
allow connections with some of the underlying metal components of the heater with the
PCB board. The next step for the designers was to deposit and pattern the under bump
metal (UBM) layers. This consisted of a Ti/Ni/Au multilayer. Soldering balls were next
printed onto the UBM surfaces. The PCB was produced with electrical connections for
Silicon block with heater on bottom
Heat sink PCB
82
the heaters and temperature sensor. The silicon blocks were placed on top of the PCB
and soldered using temperature and pressure in a flip chip bonding technique. The PCB
was next mounted to a heat sink with electrically non-conductive epoxy. The researchers
used polypropylene tape or bonded a thin sheet of plastic to the top of the reaction vessels
with epoxy or with solvent bonding. A schematic of the process sequence is shown in
Figure 4.4 (Zou, Miao et al. 2002).
Figure 4.4 Flip chip design process used by Zou et al. for fabrication of PCR
microheaters
4.1.1 Comparison of designs
The aim of a microscale device is to complete the PCR reaction as quickly as
possible with little power consumption. Table 4.1 at the end of this section summarizes
the operating parameters of the heaters designed by the researchers discussed above. The
maximum heating rate of the device manufactured by El-Ali was approximately 50 °C/s,
with a cooling rate of about 12 °C/s at a location on the heater surface. The power
consumption of the heating device was 6.3 W at 94 °C (El-Ali, Perch-Nielsen et al.
2004). The device manufactured by Zhao achieved heating rates of 15 °C/s and cooling
83
rates of 10 °C/s at a location on the heater. This device was shown to be able to
successfully complete a PCR reaction in 15 minutes (Zhao, Cui et al. 2003). Zou’s group
fabricated a device capable of a 42 °C/s heating rate and a 27 °C/s cooling rate with a 15
µL sample loaded in the reaction volume measured at the surface of the heater. Zou’s
device used, on average, only about 0.9 W per chamber . Zou’s device was capable of
successfully completing a PCR reaction with a 320 bp piece of plasmid DNA in a time of
15 minutes (Zou, Miao et al. 2002).
The manufacturing approach taken by El-Ali et al utilized SU-8 as the reaction
volume. SU-8 is a polymer. El-Ali found that the reaction volume needed to be silanized
in order to be PCR compatible. It was also necessary for Zhao’s group to add an oxide
layer to the silicon to make it PCR friendly. The reaction chamber used by Zou’s group
was made of an unspecified form of plastic and was formed with thermal molding. Zou
was the only researcher who did not report the need to passivate the surface of the
reaction volume.
Zou chose aluminum as the metal for his heater. The choice of aluminum is an
interesting one. The other two researchers presented chose platinum as the metal for the
heaters. Zou reported that he was using aluminum thicknesses upward of 2.5 µm. Zou
did not report how this was done
Zou’s device was the most ready for mass production. Zou incorporated an array
of heaters in his design and had a thermal molding process for producing reaction
chambers. Thermal molding is faster than building reaction vessels out of SU-8 or
etching them out of silicon, processes used by El-Ali and Zhao. Zou also utilized flip
chip bonding, which enabled him to quickly attach his heater assembly to the PCB board.
84
Zou’s device was the only one to utilize separate reaction chambers from the
heaters. El-Ali had his reaction volume attached to the heater and Zhao had his heater
manufactured on the same wafer as the reaction volume. This strategy by El-Ali and
Zhao was a poor one. Their heating and cooling rates were no greater than Zou’s. Zou’s
heater may be reused, however, by simply throwing away the reaction volumes and
reusing the heaters. Metal deposition is an expensive and time consuming process.
Platinum prices are approximately $40/gram so it is not a good idea to throw heaters
made of platinum away. Zou had a good strategy when considering the reuse of his
mechanism.
A summary of the designs is shown in Table 4.1. The two conventional choices
for PCR, the Peltier device, Techne Techgene studied earlier, and the LightCycler, a
convection device, use large amounts of power and have heating and cooling rates slower
than microheater systems. Neither of the traditional devices require surface treatments.
Two of the microheater devices required surface treatments due to the higher surface area
to volume ratio of these devices.
Table 4.1 Comparison of thermocycling devices (Zou, Miao et al. 2002; Zhao, Cui et al.
2003; El-Ali, Perch-Nielsen et al. 2004; LightCycler 2005)
Power Consumption
(W)
Heating Rate (°C/s)
Cooling Rate (°C/s)
Number of Reaction
Chambers
Reaction Volume
(µL)
Surface Treatment Necessary for
Biocompatibility?
Time to Complete
PCR Reaction (minutes)
Traditional Peltier Device 264 (max)
3.6 (max heating block)
2 (max heating block) 20 20 No 150
Traditional Convective Device 800 (max)
5.5 (average in capillary)
5 (average in capillary) 20 10 No 30
El-Ali's Device 6.4 (@ 94 °C)
50 (max on heater)
30 (max on heater) 1 20 Yes N/A
Zhao's Device N/A
15 (max on heater)
10 (max on heater) 1 2 Yes 15
Zou's Device 0.9 (Average)42 (max on
heater)27 (max on
heater) 1 20 No 15
85
4.2 Microheater design
There are several key aspects in designing a microheater for PCR. The first items
to be kept in mind are what will the power source be, how much current will the control
system be able to handle, how big should the heaters be, what material should the heaters
be made of, on what substrate will the heaters be fabricated on, and what heat flux is
desired.
The power required by the heater in our case was an important question. The
heater that we decided to design was to have platinum heating elements along with a
platinum temperature sensor. The heater needed to be powerful enough to increase the
sample temperature by greater than 6 K/s in order for it to heat faster than commercially
available devices such as the LightCycler which is capable of heating the reaction volume
at 6 K/s (LightCycler 2005). A 2-D finite element heat transfer package was used to
understand the power requirements needed by a microheater. A 20 W/cm2 heater was
desired, so this power requirement was examined using the 2-D package. The material
that looked promising for a reaction vessel was polycarbonate. Polycarbonate can be
easily machined, molded, and bonded (Martin, Matson et al. 1998; Klintberg, Svedberg et
al. 2003). The first case modeled was a polycarbonate well machined into a 2 mm thick
substrate. The well was 3.2 mm in diameter and 1.5 mm deep. The model was arranged
such that 10 µL of water was in the well. This meant that the height of the water in the
well was 1.2 mm. The model was axisymmetric and was done with a cylindrical
coordinate scheme. The heat flux across the centerline was considered to be zero. A
schematic of the model is shown in Figure 4.5.
86
Figure 4.5 Schematic of 2-D model used in analysis of heating power
The model was run to determine the heating characteristics of the fluid in the chamber
assuming that the water in the chamber is stationary. The results of that analysis are
shown in Figure 4.6.
0
50
100
150
200
250
300
350
400
0 1 2 3 4 5
Time (seconds)
Tem
pera
ture
(°C
)
Bottom of water at midlineCenter of water at midlineTop of water at midline
Figure 4.6 Response of 3.2 mm diameter polycarbonate chamber of thickness 2 mm
with 10 µL of water in it and a heat flux of 20 W/cm2
87
The three curves in Figure 4.6 are for three different positions in the reaction
vessel. Figure 4.5 shows the reaction vessel with the midline at the center. The three
lines in Figure 4.6 represent three positions along that axis in the water. The portion of
the chamber in contact with the water heats up rapidly while the heating rate decreases as
the distance from the heat flux increases. The heat flux of 20 W/cm2 heats the bottom
portion of the chamber to 88 °C in one second. This is satisfactory, but it is desired to
heat the water in the reaction vessel rapidly also. This model has a polycarbonate
thickness of 500 µm between the heat flux and the water with a water thickness of 1500
µm on top of it.
A polycarbonate chamber fashioned out of 2.0 mm section with a 1.5 mm deep
hole and a diameter of 5.0 mm was analyzed with the 2-D finite element package to
compare the performance with the previous example. This geometry allowed 10 µL of
water to occupy the reaction chamber at a height of 500 µm as opposed to 1500 µm in the
previous case. A schematic of the model is very similar to Figure 4.5 except that the
dimensions are slightly different. The results of the analysis are shown in Figure 4.7.
88
0
50
100
150
200
250
300
350
400
0 1 2 3 4 5Time (seconds)
Tem
pera
ture
(°C
)Bottom of water at midlineCenter of water at midlineTop of water at midline
Figure 4.7 Response of 5 mm diameter polycarbonate chamber of depth 2 mm with a 10
µL sample of water in it and a heat flux of 20 W/cm2
The average heating rate for the water at the top of the water sample for the first second is
10 °C/s and increases to 28.0 °C/s over a two second duration. These heating rates for
the sample were considered acceptable and are comparable to those achieved by the
researchers in section 4.1. It is clear that the bottom of the reaction chamber reached over
150 °C within two seconds of heating with this heat flux. This excessive heat is a
problem that will be handled by the control system. A short burst of high heat flux could
be supplied by the heater and after this the heater power could be reduced. As a result of
this modeling, a heater flux of 20 W/cm2 was chosen.
The heater flux may be related to the size of the heater and the resistance of the
heater. The power output by electricity flowing through a resistor is given by
89
IVP = (4.1)
The current is given by Ohm’s Law
RVI = (4.2)
Substituting Equation 4.2 into Equation 4.1 gives an expression relating the heater
resistance, the applied voltage, and the power.
RVP
2
= (4.3)
Now the desired resistance of the heater may be found, knowing the desired power and
the voltage. The desired input voltage was 10 V based on safety and practical
considerations of available power supplies. Two square heaters were desired, a 5 mm
and a 2.5 mm heater. The power dissipated by these two heaters considering that a 20
W/cm2 heating flux was desired was 5 and 1.25 W respectively. This gives values of 80
Ω for the 2.5 mm square heater and a resistance of 20 Ω for the 5 mm square heater.
The next step in the design of a platinum resistance heater is to determine the
heating element orientation such that the heating elements fit well within the allotted
space and the resistance requirement is satisfied. In order to design the heater elements,
the resistivity of deposited platinum must be known. The resistance of a conductor is
given by the following formula
wtLR ρ
= (4.4)
where ρ is the resistivity of the material (typical units are Ω-m), L is the length of the
wire, and w and t are the width and thickness of the wire respectively, corresponding to
the wire cross section. It was a given that e-beam evaporation followed by lift-off would
be used for the process sequence since there was uncertainty about obtaining a platinum
90
target for sputtering. Platinum heaters were initially manufactured using a resistivity
value for platinum of 1 x 10-7 Ω-m (Winter 2005). This resistivity is a bulk value for
platinum. It was discovered that platinum deposited on top of a thin chrome adhesion
layer had a higher resistivity than that listed above after experimentation. The actual
resistivity of a platinum wire of thickness 0.1 µm, a width of 50 µm and a length of 4.7
mm on a chrome adhesion layer of 150 Ǻ was 4.4 x 10-7 Ω-m as measured with a
multimeter without any heat treatment of the film. This value of resistivity was therefore
used in subsequent design calculations. Researchers have reported that this effect is due
to a large number of open and closed voids in the platinum film causing the density of the
platinum to decrease with a consequent increase in resistivity (Lourenco, Serra et al.
1998; Giani, Mailly et al. 2002).
The information for the resistivity and power requirements of the heater was used
to design two square heaters of size 2.5 and 5 mm. It was decided that the platinum
thickness would be 0.1 µm since thicker platinum layers have higher stress in them that
may cause them to crack (Rottmayer and Hoffman 1971). The 5 mm heater was designed
with 39 individual 5 mm long heater elements in parallel. The elements had a cross
section of 0.1 x 30 µm. This gave an element resistance of 725 Ω and a heater resistance
of 18.6 Ω (725 Ω/39). Masks were designed for fabrication of the heaters. Schematics of
the 5 mm heater are shown in Figures 4.8 and 4.9.
91
Figure 4.8 Schematic of 5 mm heater. All dimensions in µm.
The 39 parallel heater elements are visible in Figure 4.8. A platinum temperature sensor
is located in the center of the device. The intent of the design was to deposit a 150 Ǻ
chrome adhesion layer followed by 1000 Ǻ of platinum. The large connectors at each
end of the heater have a chrome adhesion layer deposited on top of them followed by a
layer of gold. Gold was chosen as the material for the leads because of its low resistivity
of 0.22 x 10-7 Ω-m. Low resistivity and large cross sections were designed to keep power
dissipation in this area to a minimum. The leads for the 5 mm heater were 900 µm wide
as shown in Figure 4.8. The resistance of the lead at the top of the heater was 4.5 Ω
considering that it is 9.4 mm long, 900 µm wide, and has a thickness of 0.1 µm. The
temperature sensor is clearly visible in Figure 4.9. The temperature sensor was designed
with a resistance of 617 Ω.
92
The 2.5 mm heater was designed with a heater resistance of 80.6 Ω. There were 9
heater elements each with a resistance of 725.4 Ω considering a resistivity for platinum of
4.4 X10-7 Ω-m, a line width of 30 µm, and a deposition thickness of 0.1 µm. Schematics
of the 2.5 mm heater are shown in Figures 4.10 and 4.11.
Figure 4.9 Exploded view of 5 mm heater with some elements deleted for clarity.
Center portion is the temperature sensor. All dimensions in µm.
93
Figure 4.10 Schematic of 2.5 mm heater. All dimensions in µm.
The 2.5 mm heater was similar to the 5 mm heater in that the design intent was to deposit
a 150 Ǻ layer of chrome followed by 0.1 µm of platinum. The gold leads would then be
formed by putting a chrome gold layer on top of the platinum in the lead areas. In this
design, the heater elements are not single lines, in this case. The elements double back
on themselves, thereby increasing the resistance. An exploded view of the heater
elements and temperature sensor are shown in Figure 4.11.
94
Figure 4.11 Exploded view of 2.5 mm heater. Center portion is the temperature sensor.
All dimensions in µm.
The temperature sensor in this case was designed with a resistance of 435 Ω. The heater
elements were staggered with the part of the heater that wraps back on itself on
alternating sides of the heater. The purpose of this was to give more uniform heating.
Masks were also created for the second step of photolithography which was the
deposition of the gold leads. These masks are shown in Figures 4.12 and Figures 4.13.
All of the masks shown are actually inverted from the actual masks. The actual masks
had the black area transparent and the surrounding area opaque.
95
Figure 4.12 Gold lead mask for 5 mm heater
Figure 4.13 Gold lead mask for 2.5 mm heater
4.2.1 Electronic control system design
The temperature of the heaters needed to be controlled in order to run the PCR
reaction. The options that were considered were PWM and a continuous controller that
utilized a Wheatstone bridge. The continuous controller configuration was chosen
because it made use of the heater element as the temperature sensor for the control
system. A Wheatstone bridge is a configuration of resistors that is commonly used to
measure strain gages. This configuration has been used by other researchers for
96
temperature control and is documented in the literature (Benammar and Maskell 1989).
A diagram of the circuit used to control the microheaters is shown in Figure 4.14.
Figure 4.14 Circuit schematic for microheater control
The main function of the control system is to feed current to the heater (Rh) causing its
temperature to be controlled at a fixed value for a given set of values of the variable
resistors and the digital potentiometer. The circuit works by amplifying the imbalance in
the bridge and supplying more current when the error is larger. The fact that the heater
element resistance follows a known function of temperature allows the heater element to
act as the sensing element as well as the final control element. The DS1267 is a digital
potentiometer that is controlled by a computer via a three wire serial interface. The
purpose of the digital potentiometer is to provide a setpoint to the control circuit. The
7805 is a power regulator. It takes the input voltage and regulates it to 5 V in order to
power the digital pot. The purpose of the L1494 operational amplifier is to amplify the
97
differential input in order to open or close the transistor. The purpose of the 10 kΩ
resistor around the transistor is to allow a small amount of current to flow through the
bridge at startup to initiate the control process.
The control system can be calibrated to control the temperature of the heater. The
resistance of the platinum heater increases with temperature. This fact can be used to
control the circuit. When the temperature of the heater has reached the proper
temperature, the voltage difference across the bridge should be zero. The voltage
difference across the bridge can be read at the input to the operational amplifier. This
information can be used to set the 0-10 kΩ resistor and the 0-1 kΩ resistor for a particular
heater. At bridge equilibrium the following expressions hold (Alciatore and Histand
2003)
02
2
41
1 =+
−+ hBB RR
RRR
R (4.5)
02
2
41
1 =+
−+ hEE RR
RRR
R (4.6)
where R1 and R2 are the resistances as shown in Figure 4.14. RhB refers to the resistance
of the platinum heater at the top of the PCR operating range, in this case 100 °C. RhE
refers to the resistance of the platinum heater at ambient temperature, 21 °C. R4B refers
to the resistance of the leg of the bridge containing the digital potentiometer when the
heater was at its upper setpoint of 100 °C, while R4E refers to the resistance of that leg at
ambient temperature. The potentiometer was set at 20 kΩ for operation of the heater at
its upper setpoint and 0 kΩ for operation at ambient temperature. R4 is given by the
following equations
98
p
SB
R
RR1
200001
14
++= (4.7)
p
SE
R
RR1
10001
14
++= (4.8)
where Rs is the resistance of the manual potentiometer in series with the parallel
combination of the digital potentiometer and Rp, the resistance in parallel with the digital
potentiometer. These equations may be solved simultaneously to obtain the values of the
manual potentiometers.
The onboard temperature sensor was used as a redundant method to determine the
temperature of the heater. The temperature sensor contact pads were hooked up to a
simple circuit that used a 500 Ω resistor and a voltage measurement, as shown in Figure
4.15, to determine the resistance of the sensor (Rt), and then the temperature could be
inferred from the calibration of resistance versus temperature.
Figure 4.15 Schematic of circuit used to measure temperature of platinum temperature
sensor
VRT
99
The voltage was measured across the platinum thermometer, Rt, as shown in Figure 4.15.
The current through the circuit could be found with the following equation
500
5R
VVI RT−= (4.9)
where VRT is the voltage across the platinum thermometer. The resistance of the
platinum thermometer could be found from Ohm’s Law knowing the current.
IVR RT
t = (4.10)
4.3 Microheater fabrication
The microheaters were fabricated using standard photolithographic methods.
Photolithography was performed using positive photoresist. After exposure, a
chlorobenzene soak and drive in period was used to form an outer layer on the surface of
the photoresist that would develop slower(Halverson, MacIntyre et al. 1982). E-beam
evaporation was used to deposit the chrome and then platinum (chrome adhesion layer)
after photolithography. Lift-off was used to form the heater pattern.
4.3.1 Microheater fabrication equipment and techniques
The masks for the heaters were designed in AutoCad and then converted to Adobe
Illustrator. The masks were designed as negatives, meaning that areas that were black in
the original would be clear in the final product and vice versa. The masks were printed
by RGM Graphics, Inc., of Bethesda Maryland. They were printed on a 100 µm thick
mylar sheet. The masks were setup to be printed on a standard sheet with dimensions
215.9 X 279.4 mm. A 3000 dpi printer was used to print the masks.
100
The wafers used in the fabrication process were 700 µm thick, 3 inch diameter
soda lime glass wafers obtained from Valley Design Corp. An optically clear substrate
was desired because some detection schemes involve shining light on the reaction volume
from one side and detecting fluorescent response on the other side. The wafers were first
rinsed with acetone, methanol, isopropanol, and finally with deionized water. Hot plates
were used to dehydrate the wafers and to harden the photoresist. The hot plates used
were Thermolyne type 1900 with Taylor 9940 temperature controllers.
The photoresist application, exposure and development required several items.
S1813 photoresist from Rohm and Haas was spun onto the wafers using a Headway
Research Inc. spinner at 4000 RPM for 40 seconds. Exposure was performed using the
masks from RGM Graphics and the Karl Suss MJB3 type 100DU030 photolithography
machine. The wafers were exposed with a UV light intensity of 8 mW/cm2 at 365 nm for
25 seconds. Chlorobenzene was obtained from JT Baker, product number 9179 and was
used to profile the photoresist. An oven was used to drive in the chlorobenzene. The
oven used was from Blue M, model OV-8A. Development of the photoresist took place
in a 3:1 solution of AZ400K:water. The AZ400K photoresist developer was obtained
from Microposit.
Several components were necessary for the metal deposition. A buffered oxide
etch (BOE) was necessary to prepare the surface of the glass. The buffered oxide etch
(BOE) used was BOE Improved supplied by Transene Company, Inc. The BOE was
carried out in a plastic container using plastic wafer holders. The chrome used was stock
chrome while the gold and platinum were obtained from Kurt J. Lesker Company. The
gold and platinum were 99.99% pure and came in 6.35 mm diameter 6.35 mm long
101
cylinders. The thermal evaporator used to deposit the gold was from Metra Inc., model
TEBC-22-26. The e-beam evaporation unit was a Temescal, model 1800. Profiles of
metallized features were taken with a Tencor Alphastep profilometer.
4.3.2 Microheater fabrication sequence
A simplified process sequence is shown in Figure 4.16.
Figure 4.16 Simplified fabrication sequence for microheaters
The wafers were cleaned with acetone, methanol, isopropanol, and DI water. The
wafers were next placed on a hot plate at 120 °C for two minutes to dehydrate the wafer
to prepare it for photoresist application, shown in step 1 in Figure 4.16. A 2 minute soft
bake was done at 120 °C after the application of the photoresist to prepare it for exposure.
The wafer was then exposed and soaked in chlorobenzene for 15 minutes, corresponding
to steps 2 and 3 above. The purpose of the chlorobenzene is to form an undercut
102
structure in the photoresist. The chlorobenzene penetrates into the photoresist to a depth
determined by the length of exposure to it. It removes residual solvent and low molecular
weight resins. This causes the affected region to develop more slowly then the region
underneath it (Halverson, MacIntyre et al. 1982). This aids in lift off by forming a
discontinuous metal structure that can be easily lifted off. A blown up picture of
chlorobenzene treated photoresist profile after development corresponding to step 4 is
shown in Figure 4.17.
Figure 4.17 Effect of chlorobenzene on photoresist development
The chlorobenzene soaked wafer was taken out of the solution, blow dried with N2, and
inserted into the oven at 90 °C for 15 minutes to drive the chlorobenzene into the
photoresist. The wafer was then developed in the AZ400K solution and optically
inspected under the microscope to ensure that complete development had taken place.
The wafer was metallized with chrome and platinum. The wafer was hard baked
at 120 °C for 5 minutes to strengthen the resist matrix prior to metallization. A 1 minute
BOE was performed to roughen up the exposed glass surface, allowing the chrome to
adhere better to the surface, corresponding to step 5 in Figure 4.16. The wafer was
placed into the e-beam evaporation chamber and 150 Å of chrome was deposited
followed by 1000 Å of platinum. The platinum and chrome deposition rates were kept
below 5 Å/s. The wafer was removed from the e-beam evaporator and placed in an
103
acetone bath to lift off the unwanted metal and photoresist. The wafer was soaked for
approximately 1 hour. The wafer was then removed, rinsed with DI water, blow dried
with N2, and optically inspected corresponding to step 7 in Figure 4.16.
The gold leads were then deposited on the heaters. The wafer was cleaned, blow
dried, dehydrated, and had photoresist applied to the metallized surface. A 2 minute soft
bake was done at 120 °C as in the previous step. The trace mask was then aligned to the
metallized area using the features as guides. Exposure was carried out on the wafer and
the chlorobenzene soak and drive in step were carried out. Development proceeded as
described above. The wafer was then hard baked at 120 °C for 5 minutes. The wafer was
placed in the thermal evaporator and 150 Å of chrome was deposited followed by 1000 Å
of gold. The wafer was removed from the thermal evaporator and soaked in an acetone
bath for 1 hour followed by rinsing with DI water. The wafer was then blow dried with
N2 and optically inspected under the microscope.
4.3.3 Microheater fabrication results
The microheaters suffered from some delamination problems after liftoff in the
acetone solution. Two 2.5 mm square heaters and two 5 mm square heaters were
manufactured on a wafer. The 2.5 mm heaters did not turn out at all due to a mask
problem that made it impossible to align both the 2.5 and the 5 mm heaters. 5 mm
heaters were the only ones produced for this reason. Typical delamination of heater
elements is shown in Figure 4.18.
104
Figure 4.18 Delamination of heater elements from soda lime glass substrate
The delamination shown in Figure 4.18 is for a heater with a 150 Å titanium adhesion
layer with 1000 Å of platinum deposited on top of it. This type of delamination was
common for the case where a chrome adhesion layer was used also. It was typical that
only one heater would turn out with no discontinuities in the heater elements due to
delamination out of four heaters manufactured. A discontinuity in a heater element refers
to a heater element with a section missing due to delamination or photolithography
problems. Even heaters that did not suffer from discontinuities in any heater elements
had imperfections as shown in Figure 4.19.
105
Figure 4.19 Heater elements from a heater with no discontinuities
The image shown in Figure 4.19 is from a heater with no discontinuities in any of the
heating elements. It is clear that the heater elements have pits, particularly at the edges of
the elements. The feature size of the heater elements was 30 µm. Elements with feature
sizes greater than 30 µm had fewer defects. The comparison of 50 µm wide temperature
sensor elements with 30 µm wide heater elements is shown in Figure 4.20.
Figure 4.20 Comparison of 30 µm heater elements with 50 µm temperature sensor
Heater elements
Temperature sensor
106
The 50 µm wide heater element clearly has fewer defects than the heater elements. The
result of this problem could be found in measuring the resistance of the temperature
sensor and the heater and comparing it with the design resistance. The design resistance
of the 5 mm platinum heater was 20 Ω while the design resistance of the temperature
sensor was 617 Ω. The actual resistance of the heater immediately after fabrication was
18.2 Ω and the temperature sensor resistance was 470 Ω. The design resistance was
calculated considering that the resistivity of platinum was 4.4 x 10-7 Ω-m with a
deposition thickness of 1000 Å. The actual heaters produced had a platinum deposition
thickness of 817 Å with a chrome adhesion layer underneath of it. Profilometry revealed
that this was indeed the case. Observation under a microscope indicated that the width of
the elements matched the design intent. The effective resistivity of the platinum
considering these factors was 3.6 x 10-7 Ω-m for the heater itself and 2.7 x 10-7 Ω-m for
the temperature sensor. The fact that temperature sensor resistivity was much lower than
the heater resistivity was attributed to the imperfections in the heater elements which
impede current flow through the device.
4.4 Microheater heat treatment and calibration
It was found that the resistance of the heaters and temperature sensors were
changing with time making it very difficult to control the heaters and making temperature
measurements with the platinum thermometer problematic. A study was done to
determine the effect of heat treatment on the heaters in an effort to stabilize the heater and
temperature sensor resistance. The resistance of four 5 mm heaters was tracked during
this study. The test consisted of baking the heaters in an oven at temperatures ranging
107
from 200 to 265 °C. The resistances of the heaters and temperature sensors were
measured using a multimeter. The resistances were measured at the points indicated in
Figure 4.21.
Figure 4.21 Points of resistance measurement for 5 mm microheater
The resistance of the thermometer was measured between the thermometer contact pads.
The heater consists of two heater sections in parallel if the heater ground pads are linked
together. R1 was measured between the bottom right contact pads while R2 was
measured between the bottom right contact pad and the left most contact pad. The
resistance of the heater is the parallel combination of R1 and R2 and is given as Rh. The
results for one of the heaters are summarized in the following table.
Table 4.2 Result of heat treatment on heater and temperature probe
Temperature (°C) Time Total elapsed time (h) R1 (Ω) R2 (Ω) Rh (Ω) Rt (Ω)21 0.0 0.0 N/A N/A 18.2 470
200 1.5 1.5 N/A N/A 16.7 387225 1.5 3.0 32.0 35.3 16.8 380225 0.8 3.8 31.9 35.4 16.8 376225 1.5 5.3 32.0 35.6 16.8 375265 1.0 6.3 32.2 35.6 16.9 371265 1.0 7.3 32.4 35.6 17.0 370265 1.0 8.3 32.6 36.0 17.1 369265 1.0 9.3 32.5 36.0 17.1 367225 1.5 10.8 32.4 36.5 17.2 366205 3.0 13.8 32.4 35.8 17.0 366
Heater ground pad
Heater ground pad
108
The results of the heat treatment on the 5 mm heater indicate that the resistance of
the temperature probe decreased by 22 % and the resistance of the heater decreased by
7 %. Over the last 5.5 hours of heat treatment the resistance of the heater changed by
only 0.1 Ω and the resistance of the thermometer change by only 1 Ω indicating that the
resistances had stabilized for heat treatment at these temperatures. The effective
resistivity of the heater after heat treatment was 3.2 x 10-7 Ω-m while the effective
resistivity for the temperature sensor was 2.1 x 10-7 Ω-m. It has been reported that after
heat treatment up to a temperature of 1000 °C the resistivity of a 110 µm thick platinum
film on a titanium adhesion layer reached 1.4 x 10-7 Ω-m (Lourenco, Serra et al. 1998).
The resistance of the heater was measured at ambient and at 100 °C in an oven in
order to calibrate the electronic control system. The heater used for this test was another
one of the heaters that had been heat treated but is not the same one listed in Table 4.2.
These temperatures define the limits of operation for the device. The temperature versus
resistance data was plotted for the heater and is shown in Figure 4.22.
y = 51.737x - 1035.5R2 = 0.9963
0
20
40
60
80
100
120
20 20.2 20.4 20.6 20.8 21 21.2 21.4 21.6 21.8 22Resistance (Ω)
Tem
pera
ture
(°C
)
TemperatureLinear (Temperature)
Figure 4.22 Temperature versus resistance with linear fit for microheater
109
The resistance of the heater increased from 20.4 Ω at ambient temperature to 21.9 Ω at
100 °C, a difference of 1.5 Ω or an increase in 7.4% from its ambient value. The
calibration plot for the platinum thermometer is shown in Figure 4.23.
y = 2.4215x - 855.48R2 = 0.9992
0
20
40
60
80
100
120
360 365 370 375 380 385 390 395 400Resistance (Ω)
Tem
pera
ture
(°C
)
Figure 4.23 Temperature versus resistance graph with linear fit for microheater
temperature sensor
The resistance of the temperature sensor increases from 362.6 Ω at ambient temperature
to 394.7 Ω at 100 °C. This is a difference of 32.1 Ω or a percent increase of 8.9 %. The
slope of the calibration curve for the heater is much steeper than the slope of the curve in
Figure 4.23. The percent change in resistance for each case is nearly identical but the
resistance range over which it happens for the heater is much smaller than that for the
temperature sensor.
110
Heater recalibration was redone in the oven after 1 week of use to determine if the
resistance of the temperature sensor was remaining constant. The result of that test is
shown in Figure 4.24.
0
20
40
60
80
100
120
360 365 370 375 380 385 390 395 400
Resistance (Ω)
Tem
pera
ture
(°C
)
First calibrationtemperatureSecond calibrationtemperature
Figure 4.24 Comparison of initial calibration of temperature sensor and after 1 week of
use
The results show that the temperature sensor was slightly out of calibration after 1 week
of use, especially at lower temperatures. At temperatures higher than 60 °C there was
agreement between the first and second calibrations.
4.5 Microheater testing
The microheaters were tested for heating rate, cooling rate, and power
consumption under two conditions. The first condition was with fan cooling. The fan
was turned on during cooling from 94 to 54 °C, but was otherwise off. The second
111
condition was with the cooling fan and a heat sink placed underneath the heater to speed
cooling.
The heater circuit was powered with a Mastech HY1803D power supply. The
cooling fan mounted on top of the test fixture was a RadioShack brushless, 12 VDC
blower fan, part number 273-199 as shown in Figure 4.25.
Figure 4.25 Microheater test fixture
The circuit that was used to measure the temperature of the heater, and the blower fan
were powered by an ATC power supply, model ATC-230U. The fan was supplied with
12 V, while the temperature sensing circuit was supplied with 5 V. The variable resistor
was controlled with the Instrunet 100 data acquisition system. The temperature of the
heater, heater voltage, period, and three other thermocouple temperatures were recorded
once every second by the Instrunet 100.
The second configuration of the microheater tested was with a heat sink, obtained
from RadioShack, part number 276-1368, underneath the microheater. The purpose of
the heat sink was to determine if it would help in cooling the device from 94 to 54 °C.
The microheater test fixture with heat sink is shown in Figure 4.26.
Blower fan
112
Figure 4.26 Microheater test fixture with microheater mounted on heat sink
The power calculations were made by measuring the voltage across the heater
using an inverting operational amplifier that would multiply the input voltage by -0.22
and then output that voltage to the data acquisition unit. The resistance of the heater
could be determined by knowing the calibration curve for the heater and the temperature
of the heater. The temperature of the heater was determined by the platinum
thermometer. The power supplied to the fan was significant in comparison with the
power supplied to the heater. The power consumed by the fan was 2.4 W while the
power consumption of the heater was on the order of 0.5 W. The power consumed by the
fan was included in the average and maximum power calculations for this reason.
4.5.1 Microheater testing results
The first case examined was for the heating and cooling rates of the microheater
system with the cooling fan and no heat sink. The cooling fan was only turned on during
cooling from 94 to 54 °C. The microheater was suspended in air for this test. There was
113
no test section on the heater. The purpose of this test was to determine the performance
of the heater and control system only. The results of the test are shown in Figure 4.27.
45
50
55
60
65
70
75
80
85
90
95
100
0 5 10 15 20 25 30 35 40 45 50 55 60Time (seconds)
Tem
pera
ture
(°C
)
MicroheaterSetpoint
Figure 4.27 Heating and cooling characteristics of microheater with cooling fan, no heat
sink, 12 V input
The maximum heating rate for this configuration was 16.7 K/s while the maximum
cooling rate was 7.8 K/s. The average cooling rate was 4.0 K/s. The average heating
rates were 9.2 K/s and 3.6 K/s for the intervals between 54 and 72 °C and between 72 and
94 °C respectively. The average heating rate between 72 and 94 °C was low considering
that the maximum heating rate during this interval was 16.7 K/s. This can be attributed to
poor control of the microheater during this interval and is not a property of the heater.
The case with the cooling fan and the microheater mounted on top of an
aluminum heat sink was examined. The configuration of the setup is shown in Figures
114
4.25 and 4.26. The cooling fan was turned on only during cooling. The results of that
testing are shown in Figure 4.28.
45
50
55
60
65
70
75
80
85
90
95
100
0 5 10 15 20 25 30 35 40 45 50 55 60Time (seconds)
Tem
pera
ture
(°C
)
MicroheaterSetpoint
Figure 4.28 Heating and cooling characteristics of microheater with cooling fan, heat
sink, 12 V input
The maximum cooling rate for this scenario was 14.6 K/s while the maximum heating
rate was 13.4 K/s. The average cooling rate between 94 and 54 °C was 8 K/s. The
average heating rate between 54 and 72 °C was 9 K/s while the average heating rate
between 72 and 94 °C was 11 K/s.
The power requirements of the microheater and fan were recorded. A summary
of the results along with the heating characteristics of the microheater are shown in Table
4.3. The maximum power required by the microheater system with and without the heat
sink was 2.4 W which is the power required to run the fan during cooling. During
cooling the heater is not on. The average power over a cycle for the microheater without
115
heat sink was 0.5 W. The average power over a cycle for the case with the heat sink was
1.0 W.
Table 4.3 Comparison of microheater operation with and without heat sink
Max Heating
rate (K/s)
Max Average heating
rate (K/s)
Max Cooling
rate (K/s)
Average cooling
rate (K/s)
Max Power consumption
(W)
Average power
consumption over a cycle
(W)Microheater without heat sink 16.7 9.2 7.8 4.0 2.4 0.5Microheater with heat sink 13.4 11.0 14.6 8.0 2.4 1.0
The purpose of the heat sink was to increase the cooling rate of the microheater. The
addition of the heat sink doubled the average cooling rate and increased the maximum
cooling rate by 7.8 K/s. The heat sink reduced the maximum heating rate, however,
because some of the heat energy was traveling to the heat sink. The average heating rate
actually increased with the heat sink, however, because in the case without heat sink there
was poor control at the 94 °C setpoint. The cost of using the heat sink can be seen in the
average power consumption column, however. The microheater needed twice as much
power on average to maintain the required temperatures.
116
5. MICROCHAMBERS FOR PCR
Silicon microchambers are one option for thermal cycling on a microheater or on
another flat, heated surface. Daniel et al. designed a silicon microchamber for rapid
thermal cycling. The researchers recognized that in order to perform rapid PCR the
thermal mass of the system must be small. A schematic of the device that was fabricated
is shown in Figure 5.1 (Daniel, Iqbal et al. 1998).
Figure 5.1 Silicon reaction chamber fabricated by Daniel et al.
The reaction chamber consists of a well fabricated from silicon that is suspended by four
beams. The reaction volume was designed to hold 2 µL of PCR reaction mixture along
with 1 µL of silicone oil, used to prevent evaporation.
117
Figure 5.2 Simplified fabrication sequence used by Daniel et al.
A simplified process sequence for the
device is shown in Figure 5.2 (Daniel, Iqbal et
al. 1998). The reaction vessel was fabricated
from a 400 µm thick silicon wafer with a silicon
nitride layer on both sides. Reactive ion etching
was next used to pattern the nitride layer on the
top surface to define the areas that would be
etched with KOH. The wafer was then etched
in KOH forming the pits shown in Figure 5.2.
The central pit is the reaction chamber while the
pits on each side are to provide thermal
isolation of the reaction volume. The two
lateral pits have a web of silicon nitride above
them. The purpose of the webbing was to prevent
oil from the reaction from creeping into the
thermal isolation pits. The idea behind this was that if the webbing was small enough, it
would prevent liquid from flowing into it because of surface tension. Platinum heaters
were patterned on the bottom of the reaction volume with a lift off technique using
chrome as an adhesion layer. The reaction chamber was coated with a SiO2 layer using
plasma enhanced chemical vapor deposition. The reaction volume then had a bovine
serum albumin treatment. It was found that this surface treatment was necessary to make
the microchamber PCR friendly by blocking adsorption of PCR reactants to the chamber
walls (Daniel, Iqbal et al. 1998).
118
This reaction volume was successfully used to amplify a 260 bp DNA segment.
The protocol was 5 seconds at 94 °C, 7 seconds at 55 °C, and 5 seconds at 72 °C. The
time constant for this system was 0.4 seconds (Daniel, Iqbal et al. 1998).
Shoffner et al. have also found it necessary to perform a surface treatment on
silicon chambers in order to perform PCR. The researchers first tested the effect of
treated powder silicon on PCR in standard polypropylene PCR tubes in commercially
available thermal cyclers. This was accomplished by incubating the silicon powder at
room temperature overnight in excess silanizing agent in a test tube. The excess
silanizing agent was then removed and the powder was dried overnight in an oven at 70
°C. The powder was then rinsed with distilled, deionized water and dried overnight in
the oven again. The powder was next separated into tubes and treated with 1 mL of 10
mg/mL polymer solution. The polymer solutions were prepared with 0.1 M Tris buffer.
The following polymers were used: poly-α-alanine, poly-L-aspartic acid, polyglycine,
poly-L-Leucine, poly-DL-phenylalanine, poly-DL-tryptophan, poly-L-lysine,
polyvinylpyrrolidone, polyadenylic acid, polymaleimide or maleimide. The silicon
powder was incubated at room temperature overnight at room temperature. The excess
solution was removed and the powder was oven dried overnight at 45 °C. It was found
that untreated silicon completely inhibited the reaction. Silicon powder treated with a
silicone fluid consisting mainly of dichlorooctamethyltetrasiloxane followed by a coating
of polyadenylic acid compared well with PCR run with no silicon powder (Shoffner,
Cheng et al. 1996).
This information was used to test the efficacy of PCR in silicon microchambers
(Shoffner, Cheng et al. 1996). The microchambers tested were 14 x 17 x 0.12 mm. The
119
reaction chambers had a Pyrex glass cover anodically bonded to them. The silicon chips
were first treated with the silanizing agent for 15 minutes at room temperature and then
removed. The chip was rinsed with distilled, deionized water and dried. The PCR chip
was filled with one of the various polymer solutions listed above and allowed to sit for 1
hour at room temperature. The polymer solution was removed using a vacuum, and the
chip was rinsed, and dried. SiO2 and Si3Ni4 layers of thickness 1000 Å were also applied
to some of the chips to determine the affect of these layers (Shoffner, Cheng et al. 1996).
The surface treatments enhanced PCR in the silicon microchambers compared
with untreated silicon (Shoffner, Cheng et al. 1996). The untreated silicon
microchambers showed some inhibition of the PCR reaction while those treated with the
silanizing agent and polymer coating showed results comparable to those performed in a
polypropylene tube in a commercially available thermocycler. The results from the
silicon chips treated with the silanizing agent and the polymer coating were not
repeatable, however, and this was attributed to a non-uniform surface treatment. Another
explanation was the coating may degrade over the course of the PCR reaction. SiO2 and
Si3N4 surfaces were also tested. These surfaces had a more uniform coating. The bare
silicon surface and silicon coated with Si3N4 showed inhibition of PCR. The SiO2 coated
chips showed the best performance as compared with the polypropylene tube in the
commercial thermal cycler. The results from these chips were repeatable. (Shoffner,
Cheng et al. 1996).
Silicon chambers have been shown to be PCR compatible when coated as
discussed above. Silicon microchambers are time consuming to fabricate, however.
KOH etching of silicon progresses at 1 µm/minute (Kovacs 1998). For a 500 µm wafer it
120
would take 7 hours to etch 400 µm of the material away. It has been shown above that
these wafers need additional surface treatment in order to be PCR compatible.
Researchers have focused their attention on building PCR reaction chambers out of
materials such as polycarbonate and polypropylene for these reasons.
Yang et al. have successfully completed PCR in a polycarbonate reaction volume.
The device fabricated consisted of three layers of polycarbonate thermally bonded. Each
polycarbonate layer was 250 µm thick. The top and bottom of the chamber were made of
clear polycarbonate while the middle section was made of black polycarbonate. The
polycarbonate was cut using a C02 laser engraving system. The chamber itself was a
serpentine channel, designed to decrease dead volume and bubble trapping. A picture of
the chamber is shown in Figure 5.3 (Yang, Liu et al. 2002).
Figure 5.3 Polycarbonate chamber fabricated by Yang et al.
The channel is 1.5 mm wide, 0.25 mm high, and can hold 40 µL of PCR reaction fluid.
The holes at the end of each channel were 0.7 mm in diameter. The chambers were
bonded using a Carver Inc. thermal press with a platen temperature of 133 °C and a
pressure of 1400 psi for 2 hours. The inlet and outlet holes were sealed with double sided
adhesive tape after the PCR solution was loaded. A layer of paraffin was then put on top
of the tape followed by a polycarbonate plate. A hand press was used to press down on
121
this assembly, causing the tape to be pressed down into the filling holes, creating a strong
seal (Yang, Liu et al. 2002).
This reaction chamber was used to complete a rapid PCR reaction (Yang, Liu et
al. 2002). The chamber was cycled using a dual Peltier chip setup. The chamber was
sandwiched in between two thermoelectric coolers. PCR was completed using this setup
with a total reaction time of 30 minutes. 15 seconds was spent at each temperature
plateau in the PCR cycle. The PCR yield was comparable with that achieved with a
conventional thermal cycler (Yang, Liu et al. 2002).
The PCR reaction mixture used by these researchers was modified in order to
accommodate the large surface area to volume ratio of the chamber. Polyethylene glycol
was added to the PCR reaction buffer to increase the yield. This approach has been
shown to be effective for increasing the PCR yield in polyimide microchips (Giordano,
Ferrance et al. 2001). The PCR yield was highest when using 0.75 % polyethylene glycol
8000. Different molecular weights of polyethylene glycol are available. The 8000 refers
to the molecular weight. Bovine serum albumin was also added to the reaction volume at
a concentration of 250 µg/mL to increase the yield. It was thought that the combination
of these two treatments helped block adsorption of any of the PCR reactants to the
chamber walls (Yang, Liu et al. 2002).
Other researchers have also produced polycarbonate reaction chambers. Chen et
al. has designed a polycarbonate reaction vessel for electrokinetic flow. Three
temperature zones were maintained and the fluid was moved from temperature zone to
temperature zone as required (Chen, Musundi et al. 2005). The channel designed had a
width of 100 µm, a depth of 70 µm, and was 7.9 cm long. The volume of the reaction
122
vessel was 0.55 µL. The chamber was fabricated using a brass molding die and hot
embossing in polycarbonate. The brass mold insert was heated to 180 °C and pressed
into the polycarbonate sheet at 1000 lbf for 4.5 minutes. The cover for the chamber was
thermally bonded in an oven at 150 °C for 20 minutes. The PCR reaction was degassed
in a vacuum after loading into the chamber to reduce the formation of air bubbles. 2 %
ethylene glycol was also added into the PCR reaction mixture to increase the boiling
point of the mixture and to reduce bubble formation in the device (Chen, Musundi et al.
2005).
Bubble formation in polycarbonate chambers can be a problem as noted by the
efforts taken by Chen et al. to reduce bubble formation in the chamber. Zhao et al. also
noticed problems caused by bubbles in a silicon microchamber. It was recommended that
the reaction volume not be filled completely. A small bubble should be left in the
reaction chamber to allow for expansion of the liquid upon heating. The researchers
found that the bubble should be located at the edge of the reaction vessel. Bubbles that
arose in the middle of the reaction vessel expanded upon heating and forced liquid to the
edges of the reaction chamber. This caused a problem with temperature uniformity in the
device because the temperature sensor was located directly below the heating chamber.
If all of the fluid was pushed out into the filling capillaries, that reaction volume would
not reach the required temperature (Zhao, Cui et al. 2003).
It was necessary to use a surface treatment or to add substances to the PCR
reaction vessel to inhibit adsorption of protein to the chamber walls in the research
described above that used polycarbonate and silicon. Surface treatments were done to
silicon that made the surface hydrophilic. Polycarbonate can also be made hydrophilic.
123
An oxygen plasma etch on a polycarbonate sample of thickness 1.2 mm was shown to
dramatically decrease the water contact angle (Larsson and Derand 2001). The
researchers used a Plasma Science PSO500 reactor. A 100 standard cubic centimeters
per minute (sccm) oxygen flow rate was used with the best results coming with an RF
power of 300 W. The water contact angle after this 5 minute plasma treatment was 3 °
compared to contact angles around 70.1 ° for untreated polycarbonate. The samples were
stored in aluminum foil to test the stability of the plasma treatment in dry storage.
Samples still had contact angles around 20 ° after six months of storage wrapped in
aluminum foil. This behavior was observed for polycarbonate treated with reactive ion
etch (RIE) bias voltages of 480-600 V. Intense plasma treatments also produced changes
in the Young’s modulus of the material of up to 1 GPa (Larsson and Derand 2001).
Polycarbonate can also be treated with UV light to increase the surface
hydrophilicity (Liu, Ganser et al. 2001). A 4 W manual UV lamp was used for the
surface treatment of the polycarbonate. The polycarbonate that was treated was 1 mm
thick. The polycarbonate was exposed to UV light for variable amounts of time ranging
from 0 to 13 hr. As time increases the surface becomes increasingly hydrophilic until the
contact angle levels off after 5 hours at 22 °. The polycarbonate was able to be thermally
bonded after surface treatment and no detectable changes in mechanical properties were
present (Liu, Ganser et al. 2001).
124
5.1 Microchamber design considerations
The purpose of the reaction chamber is to distribute the reaction volume in a thin
layer and have a thin layer separating it from the microheater to ensure rapid heating. A
2-D finite element heat transfer package was used to develop a sense of what chambers
might be acceptable. The first material to be examined was polycarbonate.
Polycarbonate can be thermally bonded (Bilitewski, Genrich et al. 2003; Ye, Yin et al.
2005), is easy to cut, and can be commonly found in thin sheets of 250 µm. The
microheater that was available was 5 mm square. A chamber 5 mm square or smaller
was desired for this reason. A reaction volume 5 mm square with a depth of 250 µm
would hold 6.25 µL of PCR solution. The 2-D model that was used to model this
situation is shown in Figure 5.4. The water was assumed to be static in this calculation.
Figure 5.4 2-D heat transfer model for polycarbonate chamber. All dimensions in µm.
The boundary condition around the outside of the polycarbonate, glass, and heat sink was
free convection with a value of 5 W/m2 K. A temperature boundary condition was set on
the 5 mm long border between the glass and the polycarbonate, directly beneath the water
section, to simulate the surface of the microheater. The results of running the model with
an initial system temperature of 72 °C are shown in Figure 5.5.
125
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Time (seconds)
Tem
pera
ture
(°C
)
Upper cornerMiddle top
Figure 5.5 2-D conduction model of 5 mm square polycarbonate chamber on 5 mm
square microheater
The lower trace is for a node in the upper left hand corner of the chamber while the upper
trace is for a node in the top center of the chamber. Both of these nodes were in contact
with the water. The time constant for the top middle node was 2.5 seconds while the time
constant for the upper corner node was 4 seconds. Considering that it takes four time
constants for the reaction volume to reach 98 % of its final value, this may not be an
optimal design for rapid PCR.
The reaction volume for this setup was reduced to see the effect of a smaller
reaction volume on heating between 72 and 94 °C. The reaction chamber in this second
model was 3 mm square. The polycarbonate surrounding the reaction chamber had the
same dimensions as shown in Figure 5.4. This meant that for the 5 mm square heater 1
126
mm of heater lay outside of the perimeter of the PCR solution on each side. The volume
of this reaction chamber is 2.2 µL. The results of the FEA model are shown in Figure
5.6.
50
55
60
65
70
75
80
85
90
95
100
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20Time (seconds)
Tem
pera
ture
(°C
)
Upper cornerMiddle top
Figure 5.6 2-D conduction model of 3 mm square polycarbonate chamber on 5 mm
square microheater
The upper corner trace corresponds to a node at the upper corner of the reaction vessel in
contact with the PCR mixture, where the temperature was expected to increase slowest.
The middle top trace represents a node at the top center of the PCR reaction vessel in
contact with the solution. Nodes at the top surface of the liquid were of interest because
these points would be the slowest to respond. The temperature uniformity in the reaction
vessel is much better in this case than in the case with the 5 mm reaction chamber. The
time constant for the upper corner node was 2.5 seconds while the time constant for the
127
middle top node was 2 seconds. The time constant for the top middle node remained the
same while the time constant for the upper corner node was decreased by 2 seconds
compared to the 5 mm reaction chamber. The upper corner node showed a decrease in
time constant because of the overhang of the heater in this case.
A 3 mm square reaction chamber with the same thickness shown in Figure 5.4
made from silicon with a 5 mm square heater centered beneath it was also examined to
see the effect of a different material on the time response of the chamber. Silicon has a
thermal diffusivity of 89 x 10-6 m2/s while the thermal diffusivity of polycarbonate is 138
x 10-9 m2/s(Incropera and DeWitt 1996). The silicon chamber will heat up much faster
than the polycarbonate one. The result of that calculation is shown in Figure 5.7.
50
55
60
65
70
75
80
85
90
95
100
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1Time (seconds)
Tem
pera
ture
(°C
)
Upper cornerMiddle top
Figure 5.7 2-D conduction model of 3 mm square silicon chamber on 5 mm square
microheater
128
Comparing Figures 5.6 and 5.7, it is clear the dramatic affect changing the reaction
chamber vessel material has on the speed of heating. The time scale on Figure 5.7 is
different because of the very rapid heating that the model predicts. The time constant for
a node at the middle top of the reaction volume in contact with the liquid was 0.01
seconds while the node at the upper corner of the reaction mixture in contact with the
liquid had a time constant of 0.03 seconds. This is very small compared to the 2 second
time constant of the polycarbonate chamber of identical dimensions.
The same chamber was analyzed to determine the speed at which a glass
microchamber could be heated on the 5 mm square microheater. The thermal diffusivity
of glass is 328 x 10-9 m2/s while the thermal diffusivity of polycarbonate is 138 x 10-9
m2/s. This means that the glass will heat up faster than polycarbonate. The analysis
indicated that the time constant for a node at the upper corner of the reaction chamber
was 1.1 seconds while a node at the top middle of the reaction mixture had a time
constant of 0.9 seconds indicating a uniform reaction chamber temperature.
Glass capillaries have been used for PCR in convection thermocyclers (Loeffler,
Henke et al. 2000). Glass capillaries have also been coated with indium tin oxide which
served as both a heater and a temperature sensor. PCR was performed in ten minutes
using this device (Meldrum, Evensen et al. 2000). A heat transfer model was made of a
glass capillary on the microheater to determine the time constant of the system. A
schematic of the model is shown in Figure 5.8.
129
Figure 5.8 Schematic of 800 µm capillary on microheater
The capillary outer diameter was 800 µm with an inner diameter of 400 µm. The
boundary condition around the outside of the capillary was assumed to be 5 W/m2K
except for at the bottom of the capillary where a 200 µm segment had a temperature
boundary condition. The time constant for heating between 72 and 94 °C was 1.5
seconds for a node in the water area farthest from the heater. The time constant for this
system lies in between the 3 mm square glass chamber and the polycarbonate chamber.
The biggest difference between these chambers is in the cross section of the chamber.
The glass capillary has a reaction chamber with a 0.13 mm2 cross section area while the 3
mm square chamber has a cross sectional area of 0.75 mm2. Even though the glass
capillary reaction chamber has a smaller cross sectional area than the planar design, the
time constants for both systems are comparable because planar design utilizes the heater
surface better and minimizes thermal penetration depth.
130
The above analysis demonstrates that reaction volumes made of different
materials with different sizes have different thermal time constants. The trend shows that
smaller reaction volumes and smaller penetration depths aid in heating speed. Materials
with higher thermal conductivities, such as silicon, perform much better than those with
low thermal conductivity materials like polycarbonate and glass.
A complete PCR reaction consists of many thermal cycles with temperatures
reaching as high as 94 °C. Liquid and air both expand, so it is necessary to know by how
much they expand so the reaction volume may be filled appropriately and to understand
how they should be sealed. PCR microchambers can be sealed with a drop of mineral or
silicone oil or by bonding material over the top of the filling hole (Nagai, Murakami et al.
2000; Matsubara, Kerman et al. 2005). One possible polycarbonate reaction volume is
shown in Figure 5.9.
Figure 5.9 Polycarbonate reaction vessel thermally bonded from three 250 µm thick
sheets
Water
Air bubble
131
A reaction vessel of this type is made out of three layers of polycarbonate thermally
bonded. The top layer has the filling holes, the middle layer contains the reaction vessel,
and the bottom layer has no features and is meant to seal the reaction vessel. A reaction
vessel of this type can be sealed with oil in the filling reservoirs or by bonding additional
pieces of polycarbonate over the filling holes after filling. If the ends are sealed through
polycarbonate bonding, the reaction vessel is closed and pressure may increase in the
volume during heating.
The central reaction area of a polycarbonate chamber sealed at the ends of the
type shown in Figure 5.9 was modeled. The model assumed that the system was closed
and rigid, that the air in the reaction chamber is an ideal gas mixture of air saturated with
water vapor, and that no new air bubbles arise in the reaction mixture. The governing
equations of this model are given in Table 5.1 along with a description of the model.
Table 5.1 Model to predict pressure in sealed two-phase system
Physical Meaning Equation (written as EES code) Equation Number
Total volume sum of vapor and liquid lvtot VVV += 5.1
Air component is ideal gas TRVP
mair
vairair = 5.2
Vapor pressure of water )1,,( == xTWaterpressureP ambw 5.3
Water vapor is ideal gas TRVP
mwater
vwvwv = 5.4
Liquid volume )0,,( ===
xTTWatervolumeV
mamb
lwl 5.5
Mass of water wwvw mmm += 5.6 Total pressure sum of partial pressures airw PPP += 5.7
132
where V refers to volume, P is pressure, R is the universal gas constant divided by the
molecular weight of air or water as specified by the subscript, m is the mass, T is
temperature, and x is vapor quality. The subscripts are defined as follows: w refers to
water, wl refers to liquid water, wv is the water vapor, i is an initial value, air refers to the
air-water vapor mixture in the chamber, and tot is a total amount. Pressure and volume
are EES functions that return the pressure and specific volume of water at the indicated
temperatures and qualities.
The model was run for temperatures between ambient and 100 °C. The volumes
used correspond to a reaction vessel measuring 5 mm square with a height of 250 µm.
The reaction vessel was filled with 4.2 µL of water with a 2 µL air bubble. The initial
pressure inside the reaction vessel was 1 atm. Pressure increased in the reaction chamber
as shown in Figure 5.10.
100
125
150
175
200
225
250
20 30 40 50 60 70 80 90 100
Temperature (°C)
Pres
sure
(kPa
)
Figure 5.10 Predicted pressure rise inside closed reaction vessel with water and 33% of
the volume filled with air
133
The predicted pressure at 95 °C is 210 kPa, or a gauge pressure of 110 kPa. A typical
cyanoacrylate glue (often called super glue) has a polycarbonate bonding strength of
5000 kPa (Loctite 2000). A gauge pressure of 110 kPa would create a 2.2 N force on a
circular filling hole of diameter 2 mm. Assuming that the cyanoacrylate is applied to an
area of 9 mm2, a pressure of 244 kPa would be present in the glued region. This value is
well below the polycarbonate bonding strength for cyanoacrylate making it feasible to
use cyanoacrylate to bond the cover when considering pressure only.
This model also shows that an air bubble present in the reaction vessel should also
shrink as a result of the expanding water. A graph depicting this behavior is shown in
Figure 5.11.
0.00.10.20.30.40.50.60.70.80.91.01.11.21.31.41.51.61.71.81.92.0
20 30 40 50 60 70 80 90 100
Temperature (°C)
Bub
ble
volu
me
(mm
^3)
Figure 5.11 Predicted bubble volume inside closed reaction vessel with water and 33%
of the volume filled with air
134
5.2 Microchamber fabrication equipment and techniques
Thermally bonded polycarbonate chambers of the type shown in Figure 5.12 were
fabricated. 250 µm or 125 µm thick polycarbonate film was used. The chamber profiles
were drawn in Corel Draw and then cut into the polycarbonate using a PII-30 vinyl cutter
from a company called GCC. The cutter was unable to cut all the way through the
polycarbonate so it was necessary to complete the cutting with a razor blade. The
individual polycarbonate pieces were then rinsed with isopropanol and thermally bonded
in an oven. The polycarbonate was sandwiched between two slabs of metal coated with
Teflon. The assembly was clamped together using a spring loaded clip. The assembly
was placed in the oven at 163 °C for 15 minutes and then removed. The chamber was
removed from the clamp four minutes later. The chambers were sealed with acetone,
methyl ethyl ketone, methylene chloride, or cyanoacrylate and polycarbonate.
Figure 5.12 Polycarbonate chamber fabricated from three 250 µm thick sections. All
dimension in mm
135
This particular chamber has a 5 mm square reaction chamber with 1 mm wide filling
capillaries. The reaction volume of this chamber is 6.2 µL. Polycarbonate reaction
vessels of this type were also fabricated with a 3 mm square reaction chamber. These
chambers had a reaction volume of 2.2 µL.
Chambers with an alternative design were made by drilling into a 2 mm thick
sheet of Hyzod polycarbonate from Sheffield Plastics, Inc. A Dremel Multipro model
395 was mounted in a Craftsman model 572.530320 drill press stand. A 3 mm diameter
router bit was used to create the profiles. Some chambers were sealed by placing a drop
of mineral oil on top of them. The mineral oil used was supplied by Biomerieux, part
number 70100. A schematic of one of these chambers is shown in Figure 5.13.
Figure 5.13 Countersunk polycarbonate chambers created with router bit
136
The polycarbonate chambers shown above were designed to put the PCR reaction volume
in the lower section. The thin layer surrounding the reaction volume was intended as an
oil layer. The purpose of the oil was to prevent solution evaporation.
Polycarbonate chambers of the type shown in Figure 5.13 were made hydrophilic
using an oxygen plasma etch in a Trion Technology minilock reactive ion etcher. The
polycarbonate was first rinsed with isopropanol and deionized water. The polycarbonate
pieces were than placed on a Thermolyne type 1900 hot plate with Taylor temperature
controller, model 9940 at 110 °C for ten minutes. The RIE was set to 125 W with a
pressure of 250 mTorr and an oxygen flow rate of 100 sccm. The oil reservoir was made
hydrophobic after RIE treatment by wiping it with isopropanol.
Polycarbonate was also made hydrophilic by using a portable UV light source.
The light source was a Spectroline UV lamp, model UV-4BSW from Spectronics
Corporation. The bulb used in this device emitted UV light at 254 nm.
Glass capillaries were also used as a reaction volume that had the potential for
rapid thermal cycling. The glass capillaries had an outer diameter of 890 µm with an
inside diameter of 400 µm. The capillaries were supplied by Drummond Scientific Co.,
Broomall PA.
5.3 Microchamber testing
Polycarbonate chambers were the main focus of chamber manufacture. Two
types of polycarbonate chambers were tested. The first type was the countersunk
chamber discussed earlier. The second type was the thermally bonded polycarbonate
chamber. Different methods were used to attempt to seal the thermally bonded
137
polycarbonate. Glass capillaries were also studied to determine their effectiveness for
rapid PCR thermocycling.
5.3.1 Countersunk polycarbonate chambers
Countersunk polycarbonate chambers were tested. The testing of these chambers
was done on the thermoelectric cooler (TEC) with an oil layer on top of the water with
blue food coloring. The testing involved a simulated PCR cycle consisting of an initial
six minute heating period at 95 °C and 30 cycles of PCR with 60 seconds at each
temperature and a final elongation step of 5 minutes at 72 °C. Untreated polycarbonate
chambers were tested. The chambers were simply drilled and then tested. Untreated
drilled polycarbonate wells exhibited total evaporation of the water. Large air bubbles
formed within the liquid in the chambers and eventually, all of the reaction mixture
would evaporate. A picture of these bubbles is shown in Figure 5.14.
Figure 5.14 Bubble formation in untreated polycarbonate chambers covered with
mineral oil
138
The bubbles would first start to form in independent locations and then coalesce into one
large bubble. These bubbles seemed to displace the oil allowing the water to evaporate.
The bubbles were assumed to be from trapped air on the polycarbonate surface. It
is noted that glass vessels were tested (larger geometry) similarly and no air bubbles were
observed. One of the differences between glass and polycarbonate is that glass is
hydrophilic while polycarbonate is hydrophobic. Attempts were made to make the
polycarbonate chambers hydrophilic for this reason. The first attempts at making
polycarbonate hydrophilic were with an O2 plasma etch. Different reactive ion etch
powers were used to determine the optimal power. A picture showing the results of these
tests is shown in Figure 5.15.
Figure 5.15 2 mm thick polycarbontate after O2 plasma etch at A)100 W, B) 200 W, C)
250 W
The 100 W sample showed a decreased contact angle versus untreated polycarbonate
with no deformation of the sample. The water on the surface of the 200 W sample was
almost flat to the surface as shown in Figure 5.15. The 200 W case had slight
deformation. The 250 W case showed significant deformation. Further testing showed
that the optimal level for O2 plasma treatment of a 2 mm thick polycarbonate sample
using the RIE mentioned in the fabrication equipment section was 125 W based on
physical distortion and water contact angle.
A B C
139
O2 plasma etching was done on the countersunk polycarbonate chambers. After
treatment the oil layer reservoir surface was wiped with isopropanol to remove the
surface charge on that area. This led to a hydrophilic reaction volume and a hydrophobic
oil reservoir. The purpose of this was to isolate the water in the bottom of the chamber.
These chambers allowed a simulated PCR cycle to be run in the chambers without any
observed evaporation. Bubbles still appeared in the solution, but they did not coalesce
nor disturb the oil layer as they did in the untreated polycarbonate case.
5.3.2 Glass capillaries
Glass capillaries were used frequently for PCR in this effort because it was found
that glass gave more consistent results than polycarbonate. Glass capillaries with a 400
µm inside diameter and an outside diameter of 890 µm proved to be very effective
vessels for PCR. They could easily be filled with PCR solution and capped with oil
through capillary action. There were never any problems with evaporation with these
chambers when heating on the TEC or microheater. The time constant of glass
capillaries filled with ethylene glycol was similar to those predicted by the heat transfer
model described in Section 5.1. Ethylene glycol was used for this test because it has a
boiling point of 197 °C. This allowed the capillary to be filled without having to seal the
ends with oil. The thermal diffusivity of ethylene glycol is 93 x 10-9 m2/s compared with
145 x 10-9 m2/s for water. This means that ethylene glycol will respond to changes in
temperature slower than water. Heat transfer analysis showed that capillaries filled with
water and capillaries filled with ethylene glycol should perform similarly. A graph
showing the performance of a glass capillary on the microheater is shown in Figure 5.16.
140
253035404550556065707580859095
100
0 5 10 15 20 25 30 35 40 45 50 55 60 65
Time (seconds)
Tem
pera
ture
(°C
)Heater temperatureInside capillary temperature
Figure 5.16 PCR cycle showing 5 mm square microfabricated heater and temperature
inside a 400 µm inside diameter glass capillary filled with ethylene glycol
A thermocouple was inserted into the glass capillary filled with ethylene glycol to obtain
the inside capillary temperature. The capillary temperature matches the heater
temperature very closely during cooling. During heating from 72 to 95 °C the time
constant of the heater is 1 second while the time constant for the capillary was 3 seconds.
This is similar to the calculated time constant of 1.5 seconds obtained in the 2-D analysis
which was obtained with a temperature step input.
5.3.3 Thermally bonded polycarbonate chambers
Thermally bonded polycarbonate chambers of the type shown in Figure 5.12
consisting of three 250 µm thick layers were tested on the microheater. The first
141
chamber tested was a chamber that had a 5 x 5 mm reaction chamber. The reaction
chamber volume was 6.2 µL. The chamber was filled with ethylene glycol and the
capillary channel was cut off at a point just before the square reaction chamber so a
thermocouple could easily be inserted into the chamber. A clamp on the heater setup
held the test section securely in place. The clamp was made of polycarbonate and was
off of the heated portion of the chamber. Grease was placed in between the heater and
the polycarbonate surface to enhance heat transfer. The results of that test are shown in
Figure 5.17.
253035404550556065707580859095
100
0 5 10 15 20 25 30 35 40 45 50 55 60Time (seconds)
Tem
pera
ture
(°C
)
Heater temperature
Inside chambertemperature
Figure 5.17 PCR cycle showing 5 mm square microfabricated heater and temperature
inside a 5 mm square polycarbonate reaction volume filled with ethylene glycol
When the heater temperature was at 72 °C, the temperature inside the chamber stabilized
at 68 °C as shown in Figure 5.17. This is indicative of a thermal resistance between the
heater and the thermocouple. The time constant for heating between the 72 and 94 °C
setpoint for the reaction chamber was ten seconds, which is much larger than the 2.5 – 4
142
seconds calculated with the FEA model. This poor performance is attributed to poor
thermal contact between the chamber surface and the heater.
A chamber was tested that was made by thermally bonding three 250 µm thick
pieces of polycarbonate. This chamber only had a central reaction chamber of 3 x 3 mm,
however. The volume of the central reaction chamber was 2.2 µL. This chamber was
placed onto the microheater in the exact same way as the 5 x 5 mm chamber. The
temperature inside the chamber was recorded and is shown in Figure 5.18.
253035404550556065707580859095
100
0 5 10 15 20 25 30 35 40 45 50 55 60Time (seconds)
Tem
pera
ture
(°C
)
Heater temperature
Inside chambertemperature
Figure 5.18 Simulated PCR cycle showing 5 mm square microfabricated heater and
temperature inside a 3 mm square polycarbonate reaction volume filled with ethylene
glycol
The 3 x 3 mm polycarbonate chamber followed the heater temperature closer than the 5 x
5 mm chamber. The time constant for heating between 72 and 94 °C for the 3 x 3 mm
chamber was 3 seconds. The time constant of the heater itself was 1.5 seconds as shown
143
in Figure 5.18. This value is comparable with the 2 second time constant calculated with
the conduction analysis done with a step temperature input.
A thermally bonded microchamber was tested that had reaction chamber 3 mm
square with a 125 µm thick polycarbonate section bonded to the top and bottom of the
250 µm thick middle section. The 125 µm thick sections were in place of the 250 µm
thick sections that were bonded to the top and bottom in prior experiments. The purpose
of the experiment was to see if having a thinner polycarbonate section on the bottom of
the chamber would increase the heat transfer to the chamber. The chamber was once
again filled with ethylene glycol and a thermocouple was inserted into the reaction
chamber. The chamber was clamped to the microheater. The results of this test are
shown in Figure 5.19. The time constant for the chamber during heating from 72 to 94
°C was 3 seconds while the time constant for the heater was 1.2 seconds. This value is
equal to the time constant of the same chamber with a 250 µm thick top and bottom
section.
144
253035404550556065707580859095
100
0 5 10 15 20 25 30 35 40 45 50 55 60Time (seconds)
Tem
pera
ture
(°C
)Heater temperature
Inside chambertemperature
Figure 5.19 Simulated PCR cycle showing 5 mm square microfabricated heater and
temperature inside a 3 mm square polycarbonate reaction volume with a 125 µm thick
top and bottom section filled with ethylene glycol
A 3 x 3 mm reaction chamber 250 µm thick with a 250 µm thick polycarbonate
section bonded to the top and a 125 µm thick section bonded to the bottom was tested to
determine if this variation would perform better than the case with a 125 µm thick sheet
of polycarbonate bonded to the top and bottom. The performance of this chamber is
shown in Figure 5.20.
145
253035404550556065707580859095
100
0 5 10 15 20 25 30 35 40 45 50 55 60
Time (seconds)
Tem
pera
ture
(°C
)
Heater temperature
Inside chambertemperature
Figure 5.20 Simulated PCR cycle showing 5 mm square microfabricated heater and
temperature inside a 3 mm square polycarbonate reaction volume with a 125 µm thick
bottom section and a 250 µm thick top section filled with ethylene glycol
The time constant for the reaction chamber during heating between 72 and 94 °C for this
case was 2 seconds. The performance of this chamber is superior to any of the chambers
tested above. This is clearly seen by comparing the graphs from all of the above analysis
and noting the correlation between the heater temperature and the temperature inside the
chamber. The microheater showed poor control at the 72 °C setpoint in this case.
5.3.3.1 Thermally bonded polycarbonate chamber sealing
A thermally bonded polycarbonate chamber with a 3 x 3 mm reaction chamber
with no sealing on the ends was tested on the microheater to assess its behavior. This
146
chamber was filled and placed on the microheater at 95 °C. The fluid was immediately
pushed out of the central chamber and out to the filling holes.
A thermally bonded polycarbonate chamber with a 3 x 3 mm reaction chamber
was sealed with polycarbonate and cyanoacrylate. The surface of the chamber was first
cleaned with isopropanol after filling with a 5 % ethylene glycol and water solution dyed
blue. Two small pieces of 250 µm thick polycarbonate were cut and bonded to the
chamber using cyanoacrylate. The assembly was then placed into a clamp for 5 minutes
to strengthen the bond. The chamber was then placed onto the microheater at 95 °C for
600 seconds, 30 PCR temperature cycles with 20 seconds at each temperature, and 300
seconds at 72 °C. The fluid in the central reaction volume was not immediately expelled.
An air bubble did form in the middle of the reaction volume however, and did grow. The
air bubble initially in the chamber occupied 20 % of the volume and was located on the
edge of the chamber. The filling capillaries were completely full for this test. The
bubble appeared to grow over the course of the test and upon completion occupied 30 %
of the volume and was centrally located. This was problematic because the goal is to
keep the PCR solution in the PCR chamber
One of the 3 x 3 mm square polycarbonate reaction chambers was inserted into a
vacuum chamber for ten hours prior to filling to assess the effect on bubble formation in
the chamber. The chamber was filled with water and sealed with cyanoacrylate. A small
air bubble existed in the reaction chamber prior to thermal cycling. The capillaries
leading up to the reaction chamber were filled only halfway. The fluid in the reaction
chamber was immediately pushed out into the filling capillaries upon placing on the
microheater. The difference between this test and the prior one is that the capillaries in
147
this case were only partially filled while the capillaries in the above case were completely
full. This indicates that it is important for this configuration to have the filling capillaries
completely full.
Another polycarbonate chamber that was not placed into the vacuum chamber
was filled with water. The reaction chamber was filled such that only a small bubble
existed in the reaction chamber. The filling capillaries were completely filled with water.
The filling holes were then sealed with polycarbonate/cyanoacrylate and the chamber was
placed onto the microheater. The fluid in the reaction chamber was not immediately
expelled. The chamber was heated at 95 °C for 10 minutes, cycled with 20 second time
intervals 30 times, and then heated at 72 °C for 5 minutes. The fluid volume in the
reaction chamber decreased during heating at 95 °C while the vapor bubble expanded.
During cooling the opposite happened. The volume of fluid in the central reaction
chamber appeared to be the same as before thermal cycling after the test. This test
showed that a completely full polycarbonate chamber with water could be sealed with
polycarbonate and cyanoacrylate.
A polycarbonate chamber of the same type was filled with ethylene glycol to
determine the mechanism of bubble formation. Ethylene glycol has a much higher
boiling point then water. If the bubbles in the reaction chamber were due to air trapped
between the fluid and the polycarbonate, then they should also appear in the ethylene
glycol. The chamber was placed onto the microheater at 95 °C. The air bubbles in the
reaction chamber did appear.
The results of these tests indicate that trapped air on the polycarbonate surface is
liberated during heating. The air forms bubbles that tend to push the fluid in the reaction
148
chamber. These air bubbles exert a pressure in addition to the pressure created by the air
bubble already present in the chamber and the expansion of the fluid. This can be
detrimental because the goal is to keep the reaction volume in a stationary location. The
best case scenario was for a polycarbonate chamber that had the filling capillaries
completely full and a small bubble in the reaction chamber. Glass capillaries do not
exhibit this behavior perhaps because of their hydrophilic nature. The water is attracted
to the surface and hence expels the air from the capillary upon filling. Fluid placed in
them will stay in them during heating as long as they are sealed with mineral oil.
149
6. SUMMARY, CONCLUSIONS, AND FUTURE WORK
6.1 Summary
The primary objective of this work was to investigate alternative system designs
to achieve high speed PCR. A secondary objective was to minimize power consumption.
Each of the design alternatives tested had a unique set of features that may be useful for a
particular application. Since the focus was on speed and power consumption, much of
the description focuses on these parameters. A summary comparison of all of the designs
is given in Table 6.1. The microheater was found to yield the fastest cycling with the
lowest power consumption, but due to a series of problems with vessel design, definitive
PCR results were not obtained in this configuration.
Table 6.1 Comparison of thermocyclers
Heater configurationMax heating rate
(K/s)
Average heating rate
(K/s)
Max cooling
rate (K/s)
Average cooling rate
(K/s)
Max power consumption
(W)
Average power over a cycle
(W)TEC with cooling fan 5.2 1.8 6.6 5.8 23.7 11.3TEC without cooling fan 5.8 2 5.2 4.3 23.7 7.9Techne themocycler 3.6 2.8 2.0 1.4 264 145Single fan convective thermocycler 2.0 1.2 1.8 1.2 604 255Double fan convective thermocycler 4.2 2.6 7.9 5.3 1371 645Microheater without heat sink 16.7 9.2 7.8 4.0 2.4 0.5Microheater with heat sink 13.4 11.0 14.6 8.0 2.4 1.0
A listing of the major tasks and accomplishments of the work is given in Sections
6.1.1-6.1.3.
150
6.1.1 Testing of Peltier thermocyclers
• An independent thermoelectric cooler was tested. Two configurations of this
device were tested. The two configurations tested differed in whether the fan on
the heat sink was on or off.
• The heat flux produced by the TEC was modeled using the lumped capacitance
method. The model was verified by putting the heat flux results into a 2-D heat
conduction program.
• A commercially available Peltier thermocycler was tested to assess the heating
and cooling rates and power consumption.
• The commercial Peltier device was used to study nucleotide insertion rate and
PCR in glass capillaries.
6.1.2 Testing of convective thermocyclers
• Two different convective thermocyclers were tested. One of the convective
thermocyclers tested had a single fan. The second convective thermocycler tested
had two fans. One of the fans was responsible for flowing air over the heated coil
in heating mode. The second fan was used to introduce a cool, high velocity air
stream into the vent tube.
• Placing mixing walls into the vent tube and wrapping the outside with insulation
provided the most uniform temperature within the vent tube.
• The air velocity in the convective thermocyclers was determined using a control
volume approach.
151
• The lumped capacitance method and a 2-D heat transfer package were used to
model the behavior of a polypropylene PCR reaction tube with 10 µL of water in
it in the convective thermocycler.
• The lumped capacitance model was used to investigate the behavior of reaction
vessels of different geometries and materials in the convective thermal cycler.
The effect of different air velocities was also examined.
6.1.3 Testing of a microheater thermocycler
• Microheaters were designed and fabricated based on the power requirements
necessary to heat a reaction volume greater than 6 K/s.
• A Wheatstone bridge control system was designed and tested for the microheater.
• Heat treatment was done on the microheater to stabilize the resistance.
• The microheaters were tested to assess the performance under two different
conditions. The first condition was with fan cooling only with the heater
suspended in air. The second condition was with the heater mounted on top of an
aluminum heat sink and with fan cooling.
• 2-D heat transfer software was used to analyze the performance of thermally
bonded polycarbonate chambers on the 5 mm square microheater filled with
water.
• The pressure rise inside a sealed 5 mm square rigid chamber of depth 250 µm
filled 67 % with water and 33 % with air was modeled during heating from
ambient to 94 °C.
152
• Two different types of polycarbonate reaction vessels were tested to determine
their efficacy for the PCR reaction. One type of polycarbonate reaction vessel
consisted of three sheets of polycarbonate thermally bonded. The other type of
polycarbonate chamber was fabricated by drilling a hole into a 2 mm thick sheet
of polycarbonate. Some of these chambers were treated with an oxygen plasma
etch.
6.2 Conclusions
• PCR is influenced by the ability of the heater to heat the reaction vessel. It was
shown in Sections 2.2, 2.4, 3.2, and 4.5 that different heaters have different
heating and cooling rates. The average heating rate for the air in convective
devices or the surface of the heater for heat conduction devices varied from 1.2
K/s for the single fan convective thermocycler to 11.0 K/s for the microheater
with heat sink. The average cooling rate varied from 1.2 K/s for the convective
thermocycler to 8.0 K/s for the microheater mounted on a heat sink with fan
cooling. The air in the single fan convective thermocycler takes 33 seconds to
cool from 94 to 54 °C. The surface of the microheater can cool from 94 to 54 °C
in 5 seconds, based on the average cooling rate stated above. The single fan
convective thermocycler would spend 16 minutes cooling the air to the setpoint
over the course of a 30 cycle PCR reaction while the microheater would only
spend 2 minutes cooling to the setpoint over the course of a 30 cycle PCR
reaction. This demonstrates the effect that heating and cooling rates have on the
time to complete a PCR reaction.
153
• The results of the modeling of a constant wall thickness polypropylene capillary
in the convective thermocycler filled with water shows that the time constant
increases from 5 seconds for an 890 µm diameter capillary to 65 seconds for a
cylinder diameter of 5 mm with the same wall thickness and air velocity of 1.2
m/s as shown in Section 3.3. 2.12 mm outside diameter spherical reaction vessels
made of glass containing an equivalent volume of water and an equivalent wall
thickness to a cylindrical reaction vessel of outside diameter 890 µm and inside
diameter 400 µm were shown to have time constants 4 times larger then the
cylindrical case because of the larger surface area to volume ratio of the cylinder
as shown in Section 3.3. The analysis showed that reaction volumes of different
geometries for the convective thermocycler have different time constants. Time
constants for reaction chambers can be minimized by using capillaries with
outside diameters on the order of 900 µm. The advantage of using a convective
thermocycler is that different geometry reaction vessels may easily be placed into
them.
• Analysis done in Section 5.1 with 2-D finite element analysis showed that the
time constant for two nodes, one in the corner and one at the midline, in contact
with the reaction fluid at the top of a 5 mm square thermally bonded
polycarbonate chamber fabricated from three 250 µm thick sheets centered on the
5 mm square microheater were 4 seconds and 2.5 seconds, respectively. The time
constant for two nodes, one in the corner and one at the midline, in contact with
the reaction fluid at the top of a 3 mm square thermally bonded polycarbonate
chamber fabricated from three 250 µm thick sheets centered on the 5 mm square
154
microheater were 2.5 seconds and 2 seconds, respectively. Based on these results
it was concluded that square microchambers should have perimeters that lie inside
of the microheater for greater temperature uniformity.
• The time constant for two nodes, one in the corner and one at the midline, in
contact with the reaction fluid at the top of a 3 mm square thermally bonded
silicon chamber fabricated from three 250 µm thick sections centered on the 5
mm square microheater were 0.03 seconds and 0.01 seconds, respectively, as
analyzed in Section 5.1 with a 2-D heat transfer program. Based on this work, it
was concluded that changing the material of the reaction vessel to a material with
a higher thermal diffusivity decreased the time constant and provided for uniform
heating.
• Based on the PCR time study in Section 2.4.3 it was concluded that the time
limiting step in PCR is the polymerase nucleotide insertion rate. An 879 bp
segment of DNA could not be copied with a 25 second elongation time and a
denaturation and annealing time of 15 seconds. The 879 bp segment could be
copied with a 30 second insertion rate, however. This led to the conclusion that
the nucleotide insertion rate was 29 bp/s.
• The power consumption of an independent thermoelectric cooler used for PCR
can be decreased by 30 % not using a cooling fan on the heat sink with a loss in
average cooling of 1.5 K/s, based on the analysis done in Section 2.2. This would
amount to 1 minute of time for a 30 cycle PCR reaction. Based on this analysis,
the reduction in power consumption warrants not using the fan on the heat sink.
155
• Air heating rates for the double fan convective thermocycler built in this lab were
comparable to the heating rate of the surface of the TEC and consumed 80 times
more power on average based on the analysis done in Section 3.2. Convective
thermocyclers of the type built in this lab are not capable of low power operation
with high heating and cooling rates.
• The microheaters analyzed in Section 4.3.3 fabricated with liftoff and plastic
masks suffered delamination. Platinum elements with 30 µm feature size
deposited with a thickness of 1000 Å with a 150 Å chrome adhesion layer were
pitted, especially at the edges, while 50 µm features were more uniform. The
effective resistivity for the 30 µm elements was 3.6 x 10-7 Ω-m while the effective
resistivity for the 50 µm elements was 2.7 x 10-7 Ω-m. The bulk resistivity of
platinum is 1.0 x 10-7 Ω-m. Edge deformities and pitting influence the resistivity
of the deposited platinum.
• A 13 hour heat treatment at 250 °C reduced the resistivity of the temperature
probe from 2.7 x 10-7 Ω-m to 2.1 x 10-7 Ω-m based on the analysis in Section 4.4.
Heat treatment of platinum heaters is necessary to stabilize the resistance.
• The average heating rate for the microheater was 4 times faster then any of the
other thermal cycling devices tested as analyzed in Section 4.5. The average
cooling rate was 1.5 times faster than any of the other devices tested. The average
power for the microheater was 12 % of the power required by its nearest
competitor. Microheaters are low power, fast cycling devices.
• The cooling rate of a microheater can be increased by mounting it on a heat sink
with no change in average heating performance and double the average cooling
156
rate, based on the analysis in section 4.5.1. The average power required is
doubled from 0.5 W to 1.0 W, however. A heat sink should be used to speed
cooling of a microheater.
• The pressure rise during heating from ambient to 94 °C in a rigid, closed 5 mm
square thermally bonded reaction volume of height 250 µm filled with water and
an air bubble occupying 33% of the reaction volume approaches 100 kPa using an
ideal gas mixture model discussed in Section 5.1. Sealing of the reaction vessel is
possible with cyanoacrylate and polycarbonate.
• The pressure rise in a square reaction vessel with filling capillaries on the
microheater pushes the water out of the main reaction chamber and out into the
capillaries, discussed in Section 5.3.3. If the filling capillaries are completely full
then some of the reaction will remain in the central reaction volume if the ends of
the capillaries are securely sealed. The filling capillaries for the 5 mm square
thermally bonded polycarbonate chamber comprised of a middle section of
thickness 250 µm had a volume of 6 µL, equal to the central reaction chamber
volume. This large volume provides an area for fluid to flow out into during
heating unless they are completely full.
• Air bubble formation could be seen in a thermally bonded polycarbonate chamber
filled with ethylene glycol on the microheater at 94 °C with the capillaries sealed
with cyanoacrylate and polycarbonate, based on the observations described in
Section 5.3.3. Air bubbles present in chambers filled with water expanded. Since
ethylene glycol has a boiling point of 197 °C negligible ethylene glycol was
evaporating. The other source of air in the chamber is air trapped between the
157
liquid and the chamber surface, air trapped in the polycarbonate, and air within
the fluid.
• PCR was routinely completed in glass capillaries with only mineral oil for
sealing, discussed in section 5.3.2. One difference between glass and
polycarbonate is glass is hydrophilic. Hydrophilic glass capillaries can be sealed
with mineral oil and thermal cycled.
• Chambers drilled into polycarbonate without any surface treatment filled with
water and sealed with mineral oil did not perform well in cycling tests, while
oxygen plasma treated polycarbonate wells filled with water and sealed with
mineral oil were able to be thermal cycled without the solution evaporating based
on observations discussed in Section 5.3.1. The failure mechanism for the
untreated chambers was air bubbles rising to the surface and breaching the oil
layer. The shape of the oil-water interface in the well was significantly impacted
by the surface treatment. The treated well had a relatively uniform thickness of
oil over a largely flat water surface. Oxygen plasma treated polycarbonate
chambers sealed with mineral oil can be used for thermal cycling without
complete evaporation of the sample.
6.3 Future work
The microheaters show promise for rapid thermal cycling. The microheaters
fabricated suffered from delamination problems, especially smaller features. This
problem may have been caused by contamination of the e-beam reaction chamber during
158
evaporation. The resistance of the heaters continued to change even after heat treatment
at 250 °C for 13 hours. A longer, hotter heat treatment may be necessary.
The thermally bonded polycarbonate chambers did not perform well in holding the
solution in the central reaction chamber. Polycarbonate is an attractive choice for making
reaction vessels because it can be bonded in many different ways including tape, glue,
and thermally. It can be cut and drilled very easily. Polycarbonate chambers with valves
at the inlet and outlet to the chamber could be used to contain the reaction mixture.
Another option is to make the volume of the capillaries leading up to the reaction vessel
very small so there is no room for the reaction mixture to expand into. Very small filling
holes should be utilized to minimize the pressure on the sealing method.
A detection method also needs to be implemented. It was shown in this thesis that
it is possible to run PCR with detection reagents in the reaction volume. A detection
scheme would involve an illumination source and a fluorescence detector.
A more complete system could be designed that consists of a chip that could be
inserted into an array of microheaters. This would involve designing a structure to
encase the microheaters and a system to position the reaction chambers on the
microchambers and hold them there securely. The control system also needs to be
miniaturized. A computer data acquisition unit is currently being used to collect data and
provide control. It is possible to implement this control scheme with a microchip. The
design needs to move away from plug in power supplies and toward portable battery
power.
159
REFERENCES Accuratus. (2005). "Accuratus aluminum oxide properties 96 % aluminum oxide."
Retrieved March 6, 2005, from http://www.accuratus.com/alumox.html. Alciatore, D. G. and M. B. Histand (2003). Introduction to Mechatronics and
Measurement Systems, McGraw-Hill Higher Education. Aryee, E., R. Bailey, et al. (2005). "Detection, quantification and genotyping of Herpes
Simplex Virus in cervicovaginal secretions by real-time PCR: a cross sectional survey." Virology Journal 2(61).
Benammar, M. and W. C. Maskell (1989). "Temperature control of thick-film printed
heaters." Journal of Physical Engineering: Scientific Instrumentation 22: 933-936. Bilitewski, U., M. Genrich, et al. (2003). "Biochemical analysis with microfluidic
systems." Analytical and Bionanalytical Chemistry 377: 556-569. Cadoret, J.-C., B. Rousseau, et al. (2005). "Cyclic Nucleotides, the Photosynthetic
Apparatus and Response to a UV-B Stress in the Cyanobacterium Synechocystis sp. PCC 6803." Journal of Biological Chemistry.
Chen, J., W. Musundi, et al. (2005). "Electrokinetically Synchronized Polymerase Chain
Reaction Microchip Fabricated in Polycarbonate." Analytical Chemistry 77: 658-666.
D'Amico, M. L., V. Paiotta, et al. (2002). "A Kinetic Study of the Intercalation of
Ethidium Bromide into Poly (A) Poly (U)." Journal of Physical Chemistry B 106: 12635-12641.
Daniel, J. H., S. Iqbal, et al. (1998). "Silicon microchambers for DNA amplification."
Sensors and Actuators A 71: 81-88. Draghici, S., P. Khatri, et al. (2005). Identification of genomic signatures for the design
of assays for the detection and monitoring of anthrax threats. Pacific Symposium on Biocomputing 2005, World Scientific.
Efunda. (2005). "Efunda polymer material properties." Retrieved July 8, 2005, from
http://www.efunda.com/materials/polymers/properties/polymer_datasheet.cfm?MajorID=PC&MinorID=11.
El-Ali, J., I. R. Perch-Nielsen, et al. (2004). "Simulation and Experimental Validation of
a SU-8 Based PCR Thermocycler Chip with Integrated Heaters and Temperature Sensor." Sensors and Actuators A 110: 3-10.
160
Fermer, C., P. Nilsson, et al. (2003). "Microwave-assisted high-speed PCR." European Journal of Pharmaceutical Sciences 18: 129-132.
Giani, A., F. Mailly, et al. (2002). "Investigationof Pt/Ti bilayer on SiNx/Si substrates for
thermal sensor applications." Journal of Vacuum Science Technology A 20(1): 112-116
. Gill, P. (2005). "DNA as Evidence-The Technology of Identification." The New England
Journal of Medicine 352(26): 2669-2671. Giordano, B. C., J. Ferrance, et al. (2001). "Polymerase Chain Reaction in Polymeric
Microchips: DNA Amplification in Less than 240 Seconds." Analytical Biochemistry 291: 124-132.
Goodfellow. (2005). "Goodfellow polypropylene material information." Retrieved June
16, 2005, from http://www.goodfellow.com/csp/active/STATIC/E/Polypropylene.HTML.
Gulliksen, A., L. A. Solli, et al. (2005). "Parallel nanoliter detection of cancer markers
using polymer microchips." Lab on a Chip 5: 416-420. Halverson, R. M., M. W. MacIntyre, et al. (1982). "The Mechanism of Single-Step
Liftoff with Chlorobenzene in a Diazo-Type Resist." IBM Journal of Research and Development 26(5): 590-595.
Hindson, B. J., A. J. Makarewicz, et al. (2005). "APDS: the autonomous pathogen
detection system." Biosensors and Bioelectronics 20: 1925-1931. Incropera, F. and D. DeWitt (1996). Fundamentals of Heat and Mass Transfer, John
Wiley & Sons. Kermekchiev, M. B., A. Tzekov, et al. (2003). "Cold-sensitive mutants of Taq DNA
polymerase provide a hot start for PCR." Nucleic Acids Research 31(21): 6139-6147.
Klintberg, L., M. Svedberg, et al. (2003). "Fabrication of a paraffin actuator using hot
embossing of polycarbonate." Sensors and Actuators A 103(3): 307-316. Kopp, M. U., A. J. de Mello, et al. (1998). "Chemical Amplification: Continuous-Flow
PCR on a Chip." Science 280: 1046-1047. Kovacs, G. (1998). Micromachined Transducers Sourcebook, McGraw-Hill. Lane, S., J. Everman, et al. (2004). "Amplicon secondary structure prevents target
hybridization to oligonucleotide microarrays." Biosensors and Bioelectronics 20(4): 728-735.
161
Larsson, A. and H. Derand (2001). "Stability of Polycarbonate and Polystyrene Surfaces
after Hydrophilization with High Intensity Oxygen RF Plasma." Journal of Colloid and Interface Science 246: 214-221.
Lee, D. S., S. Ho Park, et al. (2004). "Bulk-micromachined submicroliter-volume PCR
chip with very rapid thermal response and low power consumption." Lab on a Chip 4: 401-407.
Lee, D. S., C. Y. Tsai, et al. (2004). "A new thermal cycling mechanism for effective
polymerase chain reaction in microliter volumes." Microsystem Technologies 10: 579-584.
LightCycler. (2005). "Technical Aspects of the LightCycler Instrument." Retrieved July
14, 2005, from http://www.roche-applied-science.com/sis/rtpcr/lightcycler/. Liu, Y., D. Ganser, et al. (2001). "Microfabricated Polycarbonate CE Devices for DNA
Analysis." Analytical Chemistry 73(17): 4196-4201. Liu, Y., C. Rauch, et al. (2002). "DNA Amplification and Hybridization Assays in
Integrated Plastic Monolithic Devices." Analytical Chemistry 74: 3063-6070. Loctite. (2000). "Loctite Laboratory Data Sheet, Product 190848." Retrieved August 9,
2005. Loeffler, J., N. Henke, et al. (2000). "Quantification of Fungal DNA by Using
Fluorescence Resonance Energy Transfer and the Light Cycler System." Journal of Clinical Microbiology 38(2): 586-590.
Lourenco, M. J., J. M. Serra, et al. (1998). "Thin Film Characterization for High
Temperature Applications." International Journal of Thermophysics 19(4): 1253-1265.
Martin, P. M., D. W. Matson, et al. (1998). Fabrication of plastic microfluidic
components. Conference on Microfluidics Devices and Systems, Santa Clara, CA, The International Society for Optical Engineering.
Matsubara, Y., K. Kerman, et al. (2004). "On-Chip Nanoliter-Volume Multiplex TaqMan
Polymerase Chain Reaction from a Single Copy Based on Counting Fluorescence Released Microchambers." Analytical Chemistry 76(21): 6434-6439.
Matsubara, Y., K. Kerman, et al. (2005). "Microchamber array based DNA quantification
and specific sequence detection from a single copy via PCR in nanoliter volumes." Biosensors and Bioelectronics 20: 1482-1490.
162
Melcor. (2005). "Melcor thermal solutions." Retrieved July 23, 2005, from http://www.melcor.com/pdf/MTScat.pdf.
Meldrum, D., H. Evensen, et al. (2000). "Acapella-1K, A Capillary-Based Submicroliter
Automated Fluid Handling System for Genome Analysis." Genome Research 10(1): 95-104.
Moran, M. J. and H. N. Shapiro (1999). Fundamentals of Engineering Thermodynamics,
John Wiley & Sons, Inc. Nagai, H., Y. Murakami, et al. (2001). "Development of a microchamber array for
picoliter PCR." Analytical Chemistry 73(5): 1043-1047 Nagai, H., Y. Murakami, et al. (2000). "High-throughput PCR in silicon based
microchamber array." Biosensors & Bioelectronics 16: 1015-1019. Nam, H.-M., V. Srinivasan, et al. (2005). "Application of SYBR green real-time PCR
assay for specific detection of Salmonella spp. in dairy farm environmental samples." International Journal of Food Microbiology 102: 161-171.
Nitsche, A., N. Steur, et al. (1999). "Different Real-Time PCR Formats Compared for the
Quantitative Detection of Human Cytomegalovirus DNA." Clinical Chemistry 45(11): 1932-1937.
Pyun, C.-H. and S.-M. Park (1985). Fluorescence Enhancement Mechanisms of Ethidium
Bromide by DNA or DNA Base Molecules. The Electrochemical Society Extended Abstracts Fall Meeting, Las Vegas, NV, Electrochemical Society.
Rotron, C. (2005). "Comair Rotron Patriot AC fan." Retrieved July 5, 2005, from
http://www.comairrotron.com/images/curves/patriotac3350.pdf. Rottmayer, R. E. and R. W. Hoffman (1971). "Structure and Intrinsic Stress of Platinum
Films." Journal of Vacuum Science Technology A 8(1): 151-155. Scheinert, P., B. Behrens, et al. (2005). "Optimizing DNA Amplification Protocols Using
the Eppendorf Mastercycler." Retrieved July 6, 2005, from http://www.eppendorfna.com/applications/PCR_appl_protocolsMC.asp.
Selvaganapathy, P. R., E. T. Carlen, et al. (2003). "Recent Progress in Microfluidic
Devices for Nucleic Acid and Antibody Assays." Proceedings of the IEEE 91(6): 954-975.
Shoffner, M. A., J. Cheng, et al. (1996). "Chip PCR. I. Surface passivation of
microfabricated silicon-glass chips for PCR." Nucleic Acids Research 24(2): 375-379.
163
Vierstraete, A. (2004). "PCR: Polymerase Chain Reaction." Retrieved June 1, 2005, from http://users.ugent.be/~avierstr/principles/pcr.html.
Winter, M. (2005). "Web Elements Periodic Table." Retrieved July 14, 2005, from
http://www.webelements.com. Wittwer, C. T., G. C. Fillmore, et al. (1989). "Automated polymerase chain reaction in
capillary tubes with hot air." Nucleic Acids Research 17(11): 4353-4357. Yang, J., Y. Liu, et al. (2002). "High sensitivity PCR assay in plastic micro reactors." Lab
on a Chip 2: 179-187. Ye, M.-Y., X.-F. Yin, et al. (2005). "DNA separation with low-viscosity sieving matrix
on microfabricated polycarbonate microfluidic chips." Analytical and Bionanalytical Chemistry 381: 820-827.
Yoon, D. S., Y.-S. Lee, et al. (2002). "Precise temperature control and rapid thermal
cycling in a micromachined DNA polymerase chain reaction." Journal of Micromechanics and Microengineering 12(6): 813-823.
Zhao, Z., D. Cui, et al. (2002). "An integrated biochip design and fabrication."
Proceedings of SPIE-The International Society for Optical Engineering 4936: 321-326.
Zhao, Z., Z. Cui, et al. (2003). "Monolithically integrated PCR biochip for DNA
amplification." Sensors and Actuators A 108: 162-167. Zou, Q., Y. Miao, et al. (2002). "Micro-Assembled Multi-Chamber Thermal Cycler for
Low-Cost Reaction Chip Thermal Multiplexing." Sensors and Actuators A 102: 114-121.